August 17, 2023 • 48 mins
Shanif Dhanani from New York is an AI and ChatGPT consultant with a background in software development and data science. He currently helps businesses understand and build systems that connect their internal and proprietary systems to ChatGPT, enabling internal employees, customers, and other stakeholders to reap the benefits of using large language models on their data.
He has built machine learning, analytics, and software systems at large companies like Twitter and Booz Allen, smaller startups like Apteo, where he was the founder, and TapCommerce, which was one of the world's first mobile advertising platforms.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:05):
The show you need to get what you desire byavoiding the mistakes made by others before you
learned the stories and journeys of whatsuccess looks like to find a freedom you to
serve Weiss thriving with your best zotes withAndrew Weiss.
(00:32):
I'm back to another episode of rep results withAndrew Weiss.
We have the incredible Shanif Danani here.
And in case you're wondering who this Shanifguy is, He is from New York, and he's an AI and
chat GPT consultant with the background insoftware development and data science.
He helps businesses understand and buildsystems that connect their internal and
proprietary systems to chat DBT, enablinginternal employees, customers, and other
(00:56):
stakeholders to reap the benefits of usinglarge language models on their data.
He's worked with companies like Twitter, BoozAllen, and Tap Commerce, and he now has a CEO
and founder of his own company calledLacoustive.
Did I say that right?
Locustiv.
We got it.
Locustiv.
And so welcome to the show show, Shaniv, tellus what's the biggest and best business deal
(01:17):
you're most proud of.
Oh, man.
Starting right off the bat.
Well, Andrew, thanks for having me.
First of all, it's been an awesome, awesome tobe here.
I apologize for those of you.
I'm in New York.
He might hear sirens in the background.
Oh, yeah.
Very standard for those of us here.
But, no, the the you know, what?
The biggest and coolest thing I've done, we itwas back when we started Tap Commerce, which is
maybe more than 10 years ago at this point.
(01:38):
My background is a tech and AI like you heard,and we built 1 of the world's first mobile
advertising forms at our peak, we were handlingabout a 1,000,000 ad requests a second.
I built a lot of the AI that was using, datascience to figure out how to price these ads as
well as how to respond to them at such a largescale.
So that was really cool.
Kind of the highlight of my tech career.
We eventually sold that company to Twitter, soit all sort of worked out well.
(01:59):
But that's probably the thing I'm really proudabout in terms of something I was able to
accomplish, in my career.
What does that mean a 1,000,000 ad requests asecond.
Oh, it means it means we had a lot of,hardware.
So essentially, you know, the world ofadvertising, you probably have had a a few
guests been in this world.
It is crazy how complicated it is.
(02:21):
When when mobile phones first came out, theystarted to be the this ad ecosystem where
publishers who created apps wanted to get paidby showing ads.
And so there's this ecosystem now where thosepublishers can send an ad request something
like an exchange.
The exchange then fans that request out to abunch of advertiser representatives.
The representatives then have to decide do theywanna show an ad on this product this happens,
(02:45):
you know, this is happening 1000000 of times asecond, and you have to be able to respond in
less than a 100 milliseconds, which is like,how fast you blink your probably faster.
And so it means that we had built up a lot ofsoftware and a lot of infrastructure to be able
to show an ad to the right person at the righttime for the right price.
Oh, that is so cool.
And you know, selling that company to Twitter,what was was that bittersweet, or were you,
(03:08):
like, Victoria C when they sold theircompanies.
Like, jeez, I could've taken this to the moon.
Weiss they have to buy me out like that?
You know, from a from a techie's perspective,it was kinda cool.
It was an opportunity for us see how a largercompany was doing ads at the time.
So he got to work on Twitter's ad platform.
From the perspective of, you know, I'm astartup guy at heart.
So from the perspective of going from somethingsmall to something much, much larger.
(03:33):
I I realized that the larger thing wasn't forme, which is why I went back the startup world,
but it was one of those things where we worked.
That was one of the hardest times I've worked,man.
We were working long hours, 14, 15 hour days,Andrew when you're dealing with a 1,000,000 ad
requests, a lot of your stuff crashes a lot.
And so we had to pick stuff in the middle ofthe night.
I would sometimes get get a buzzer at 3 AM, andI'd have to wake up and start stuff over again.
(03:55):
And so for me, it was a nice respite.
Good transition, but now I'm sort of back at itin the world of startups.
Oh, that that's so cool.
And, I'm very excited to talk about, chat GPTand NII with you today.
So tell us more about this overnight success.
I mean, chat GPT had, what what was it, like, a100,000,000 users in a month or something?
(04:18):
And the next closest competitor.
It was a 100,000,000 users in 2 years.
So what led us to the evolution of how chat tbtbecame so popular in the 1st place?
You know, I don't know, if you guys experiencedthis, but the first time I tried chattyBT, it
was like an moment because I'm a software guy,and a lot of my time is spent looking up stuff
(04:39):
online or trying to fix really nuanced bugs.
It take a lot of time.
And the first time I've tried soft chat JBT tofix the software.
It got me the right answer within a couple ofseconds.
And so I think that's emblematic of why thisthing has blown up.
You know, you're suddenly getting huge amountsof value.
You're saving lots of time.
Andrew you're getting a lot of actually reallygood information at your fingertips in seconds,
(04:59):
whereas it might have taken even 4 or 5 hoursto find a piece of information before,
certainly the case in software development.
Andrew so, you know, they say that everyovernight success is 10 years in the making.
Right?
So GPT 3.5, which came out earlier this year,was following 3 that followed 2, and they've
actually been working on this stuff for a longtime, lots of years.
But, 3.5 was so good that I think it just got alot of people really excited and really
(05:23):
interested, and it just sort of the hype blewup.
And it it it was really cool to see and cool tobe part of.
At least it's interesting for me to be part ofthat process now.
That's so cool.
And how did you get involved in the space?
Obviously, you know, you have a lot ofexperience in tech and e commerce and all
these, language learning models.
But how did you how were you able to dive inand say, oh, I'm a consultant for this because
(05:46):
I know exactly how to use it.
Yeah.
You know, what's funny is, I started using itAndrew then I found that it didn't do a few
things that I wanted.
I like for example, I couldn't just ask it,hey, how many leads do I have in Salesforce, or
how many what's in my product documentation,which lives on Google Drive, give me the
contents of page 33.
So like what I started seeing was a lot ofbusiness owners wanted to use this, but they
(06:10):
couldn't because a lot of their docs werebehind firewall or you needed to log in.
And by default, chatTPT doesn't have access tothese things.
And so it was kinda cool for me because mybackground is in AI and software and data.
So I was able to build the pipes that connectedyour Google Drive or your Salesforce account to
chat GPT.
And a lot of people were like, oh, that's kindacool.
Now can I do this for my, you know, GoogleSheets Andrew can I do this for my email Andrew
(06:34):
And can you add this from my Slack workspace asa chatbot?
And so a lot of folks were just starting to getinterested in what became possible by
connecting their data to chat GT Andrew becauseI had I have a lot of experience building those
sorts of data tools, I was able to jump in, andhopefully deliver something that they
appreciate.
I love that.
And that's the best way to, yeah, get good atthings is to see where the gaps are and go.
(06:58):
Well, I can fix that.
I know how to I know I know how to, adjust forthat.
So so let's talk more about to so you yourjourney with chat TBT is, you're like, oh, this
is so cool.
It can do something in seconds, which I've beenin a career would usually take hours.
It's still missing things.
I can figure out how to close those gaps.
So so tell us more about, what what do you wishyou knew earlier when you're just starting off
(07:23):
with the using chat g b t in the first place.
You know, it's funny that they're asking thatbecause we're still so early in the world of
chat g b t.
But 6 months almost feels like 10 years, right,from a few like, things move so quickly that
now a short amount of time feels like it wasthe result of a very large, large amount of
sense.
So for me, what do I wish I knew earlier thisyear?
(07:43):
I think that there's, there's still a lot ofThere's still a lot of things that's that I
struggle with when it comes to implementingchat TPT on my own data.
For example, one of the biggest limit right nowis that you can only send in a few thousand
words every time you wanna get an answer, andyou also, whenever you ask an answer, you also
need provide the right context for chat dbt toanswer you using your data.
(08:08):
I passed, with with
the chat dbt update.
I was told you could upload, like, a book, thesize of the lord of the rings, or is that not
come come through yet?
That's what they say, man, but it's 32,000tokens equates to about 8000 words, which is
maybe 20 or 30 pages, maybe 40 pages.
But Oh.
You know, essentially, you're really limited.
Right?
So the this would be awesome if I could.
(08:28):
If I could just take all one thousand pages ofmy product handbook and dump it into chat, you
could see, and ask the question about somethingthat lives on page 534.
We're good, but, but you can't.
And that's a lot of the things that businessesare running into now.
So, you know, I wish I had known that before,but in a certain sense, it's It's also provided
me with an opportunity to build the softwarethat lets people find the right, pieces of
(08:50):
their data that I can send to chat GPT so thatthey can actually ask a question.
So that's one thing.
Another thing is, hey, chat t p t if you use itthrough the API, they don't let you send a lot
of information.
They don't let you send a lot of requests.
You kind of get what's called rate limited ortimed out a lot, which means you have to sort
of take your time when you send them stuffAndrew it makes your product, my product, a
(09:11):
little bit slower for users.
Still usable by by far, but still one of thosethings I wish I had known earlier.
And there's all sorts of things like this.
Now I'm nitpicking.
Right?
This is such a cool technology.
It's all interesting.
These are really minor things that I'm sureOpenAI is gonna fix.
But the point is if you're trying to build ontop chat JPT or you're trying to use it for
things like connecting your own data, you haveto be ready for a lot of these challenges.
(09:35):
And if you're not, maybe it's better to use athird party tool.
No.
That that makes it makes a lot of sense.
So just being aware of what it can and can'tdo, it's definitely definitely good to know.
And I know you bring yourself as a chat tint,an LOM consultant, but how much have you played
around with, Bard Andrew Claude and and allthose?
I'm curious your your quick thoughts on allthose.
(09:56):
I like LAMA.
It's super fast.
I love Claude.
You know, Claude is one of those things forthose who don't know.
It's from this company called Antropic,competitor to chat JPT.
Claud actually lets you put in something like75,000 token, which is, 2 to 3 times what, you
know, what chat TPT 4 lets you do.
For what it's worth, I use chat TPT 3.5 withthe 16,000 token limit because it's much
(10:17):
cheaper.
I digress.
Cloud's awesome.
The reason it's awesome is because because youcan put in a lot more, information, and it's
super fast.
You can get sort of a lot of long form contentfrom it a little bit more quickly and a little
bit more accurately.
These LOM still aren't great for long formcontent.
They're still a little bit robotic and generic,but cloud does a really good job of providing
(10:39):
you with, like, logs or articles ornewsletters, you'll still have to edit them,
but it's a good start.
Llama is really fast.
Llama's a the open source tool from Facebook,really good Bard, I've used a little bit.
I didn't think Bard sort of got a little wonkyfor me when I tried it, but this was about a
month or 2 ago.
So I'm sure it's gotten a lot better now.
Shows you how fast things are going.
I think the whole ecosystem, the fact thatthere are so many of these providers is
(11:02):
awesome, particularly for companies that are inregulated fields like medicine because they're
gonna need to build their own open sourceversions of this or at least implement them
because they can't send data outside theirsystems.
So it's a long way of me saying, I think thesethese tools are all really cool.
What what I got out of that is it sounds likeif you have a lot of, words and, like, product
information, you have to pull from or summarizeor stuff like that, cause the way to go because
(11:28):
you can upload a lot more tokens to get moreinformation.
It sounds like chat, GPT, though.
They're still top for, like, create activitywriting and, like, being able to, like, match
your writing style per se.
Andrew sounds like llama's kind of a mixture ofthose.
Is that kind of a 3rd grade summary of things.
It's it's a good, like, yeah, it's a good highlevel approach.
You know, obviously, there are nuances, detailswith everything.
(11:50):
For so for example, a lot of people don't thinkthought is as good with software development.
So I I I probably would fall in that camp, butI still use chat JPT for a lot of my coding.
I use Claude for a lot of my long form content.
Llama, I use every now and then mostly for thenovelty.
Bard, I don't use as much unfortunately, but II'm sure I will as Google makes it better.
(12:11):
And so it's one of those things where you sortof get used to doing different things with
different tools, and you're probably gonnastick with that until then, unless another tool
comes out, that's just a lot better.
One thing to keep in mind, though, is evenClaude, which can take a lot more content, it's
still limited, and it's still limited to maybe75,000 tokens, which is about, I don't know,
20,000 words or so to give or take, 13,000.
(12:33):
You're still not gonna be able to dump a wholelot of information into it.
And so that's one of the biggest issues withbusiness today is if you've got a lot of
documentation, How can you make sure that youcan answer questions from that documentation
given these what's called context windows ortoken limits?
Interesting.
And you Andrew you keep saying that tokens, youknow, why does it have to be tokens versus
(12:53):
just, like, words or characters, essentially?
Yeah.
It's, I apologize.
When you're sort of in this field, you startusing the jargon, these tool, these companies
like OpenAI Andrew Google and and propic,they've they've introduced this concept of a
token, which is Kind of like a couple ofletters or a part of a word or an individual
(13:15):
character like maybe a semi colon.
Now the reason they use tokens instead of fullwords is because the AI that runs this stuff
basically processes the text that you get insmall chunks and what they they call these
things tokens Andrew when you have a token,it's actually a lot easier to turn those tokens
into some sort of numerical representation thatcomputers are better at dealing with than if
(13:37):
you had a whole word.
So you might hear the word token in general,one token.
They say 4 tokens is maybe a word.
I'm sorry.
They say 4 characters is about one token.
And so that's how you can start to think aboutthis stuff.
Interesting.
Okay.
That's good to know.
So let's talk about, you know, I know a lot ofpeople are still, you know, even though with a
(13:59):
100,000,000 Andrew counting and maybe it's at200,000,000 by now of users of AI, there's
still a lot of talk in the industry of Like, Idon't know if I can use AI.
It sounds too complicated for me.
What are the all the ethical implications withit?
So what do you tell people to help comfort themabout being okay with using AI and leveraging
it?
You know, the first thing I tell them is if youdon't do it, somebody else will.
(14:21):
So, you know, you're you're probably, I mean,unless you're close to retired or you don't
care anymore, then it's fine.
Right?
Who cares?
But if you're really, you know, don't know andyou're you're very career oriented, like, like
those of us in New York tend to be, youprobably wanna get the best tools for your job.
And AI is probably gonna be in that toolkit atsome point if it's not already.
(14:43):
So I would generally say I understand thatthere's some sort of hesitancy to using it, but
also you can just go and start using it rightaway, and it's very easy, and it actually is
very simple to use.
And so that's how I started.
I just went over to chat TPT's website.
They give you a few credits, you know, for freeto get started, and I just started asking it a
few questions.
(15:04):
Andrew once when I had a coding issue, I hadsaid, hey.
How do I solve this?
And it it came back with the right answer.
And then I started using this tool calledCopilot, which is a after code.
And so, like, you can start really simply.
You don't need to dive all in.
The folks who make these tools make it reallyeasy to get started, and you're only doing
yourself a disservice if you don't know whatthey are.
And so it's it really behooms you to kinda getfamiliar with these tools.
(15:26):
Now maybe you say, hey.
Look.
I tried it.
It has no bearing on my work.
It doesn't make any sense for me.
There are folks like that.
That's fine.
At least you've tried it.
I would I would argue that's probably not thecase for most people, but sure if that's your
opinion, sure.
But you gotta at least try it because it'seasy.
It's free.
It takes 2 seconds, and it probably wouldchange your life at some point.
Yeah.
And I love, the analogy of, know, sometimes,you know, AI only puts out what you put into it
(15:53):
essentially.
And so if if you, like, tell it, you know, makeme a $1,000,000.
You know, it's probably gonna be like, what?
But it, obviously, it's to keep it.
I think some person said it's kinda a atoddler, like, hey.
Can you, go collect this book from thisbookshelf and bring this Sunrise information,
or can you solve this little problem for me?
It's like, okay.
Yeah.
I got you kinda thing.
(16:14):
Andrew was following a tweet too about, like,the hustle GPT guide.
Did you hear about that at all?
No.
I haven't
so he basically was doing he started a Twittertrend.
He's like, okay.
I'm gonna see if Chad GPT can help me make a$1,000,000.
So I'm gonna name it, hustled GPT Andrew tellme what to do every day for 30 days straight.
And he essentially made all of his money offpeople joining his community.
And so technically, you you he's getting closerand closer.
(16:41):
But but just, yeah, it's it's a reminder toothat, like, you know, it it can only put out
what we put into it Andrew just, like,constantly having someone like Shanif to help
you, like, tweak it and make sure you get thebest out of it.
Sure, is definitely essential.
Andrew so speaking of that, yeah, she tellssome more examples.
Let's say you're you're a business owner andyou're like, alright.
How can I be better at marketing my business?
(17:03):
Andrew I sell, let's just say you you sellvoice, coaching lessons, essentially.
How how would you use a chat GPT to build yourvoice, voice coaching business?
You know, this is it's a really interestingquestion.
Because a lot of businesses want to use thistool, but they don't know how.
And I almost always recommend them to startwith a small problem that they have So take a
(17:27):
look at what you do on a on a day to day basis.
And if there's something that has just taken upa little bit of time, 20, 30 minutes at a time,
but you have to do it a lot.
Andrew maybe you might be able to automate it,but you probably have to do it yourself because
it requires a little bit of thought.
That's a really good opportunity to use chatJPT.
Now maybe that's not necessarily growing ormarketing your business.
Maybe that's operations, but I bet you if youfind something that you're doing a lot, that's
(17:51):
taking up a little bit of like, a good amountof time, and it's not really necessary.
It's a bit frustrating.
You probably could automate that away usingsomething like chat GPT.
Now, if you really wanted to start creating newrevenue generating opportunities, there's a few
ways you could apply this, like say, let's justbreak it down sales, marketing, and product.
On the sales side, you've got folks today whoare using chat JPT to help you write emails.
(18:16):
Lot of times, it's gonna be personalizedemails.
Like, you can copy and paste somebody'sLinkedIn profile and say, hey, chat to PT,
like, you know, create a personalized email forthis.
A lot of times, you can start doingpersonalization at scale.
There's a few tools out there right now thatlet you scrape someone's LinkedIn and then
write an email and do this several 100 or 1000times a day.
What
tools are those?
That sounds pretty valuable.
(18:37):
There's I think there's one I started lookingat called reply.i0, if I remember correctly.
I haven't used this myself.
However, it's something, yeah, it's somethingthat I wanna use at some point soon, drive your
revenue growth with Weiss So replying.i0 willlet you automatically personalize things at
scale.
As far as I know, there's a few other tools outthere, that lets you do something like this
(18:58):
similar.
I have no affiliation with the Play Datta.
I just saw them Monday.
And then on the mark, let's see.
On another way that sales folks at least arestarting to use my stuff, is a lot of times
they'll get these really obscure questions fromtheir customers on a call.
For example, like one of my customers doesfinancial retirement planning or government
employees, Andrew the government has all sortsof crazy policies.
(19:19):
So, like, the customer, theirs might say, hey.
I'm a I'm a spouse of a surviving military, youknow, benefits, receiver, I I just hit seventy
years old.
Can I start borrowing from my federalretirement account next month?
Like like, what?
Andrew so these guys are on a call.
They get a co a question like this on theircustomer.
What they were doing before was they would haveto message their team internally to look this
(19:42):
up up Andrew then they tell the customer, hey.
I'll get back to you.
Now because they have a tool like mine that canconnect their internal policy documents to a
chatbot, they can just ask the chatbot andthey'll get an answer back in a few seconds,
and they can tell the customer the answer rightaway.
And so sales guys are now using tools like thison their calls to answer customer questions.
They're using it to transcribe their calls andget a sense for what messages work and what
(20:05):
they'll So, again, it you talk about, hey.
How does chat JPT grow my business?
I like to say it's really chat JPT and othertools like this are really good at tasks.
But they suck at jobs.
So if you give them a job, like you said, tella toddler to go get my book from the shelf and
tell me what's on page 3.
It's gonna be great.
But if you say, like, you said, hey.
How do I make me a $1,000,000?
(20:26):
It's not gonna come back and tell you what todo, or maybe it is if you're this one guy.
But if you're anyone else, it's not gonna tellyou anything.
So you as the business owner need to know whatyour job is, what are your tasks in that job,
where can you start to automate some of thisstuff away, and how can you use automation
right now to start growing in those areas.
Yeah.
It's a long way of answering your question.
No.
(20:46):
And and it's great because you you mentioned,well, how how does operations grow your
business?
Mean, I would say, well, if you're able toautomate your operations, that gives you more
time for marketing, more time for fulfillment,more time for sales, and be able to keep
outsourcing time to other things that stillbecause it still takes time to grow your
business.
If you're so a brand new or small business, yougotta wear many hats where it's always better
(21:07):
to outsource and delegate those other hats.
So my next question was asking, can Andrew givemore, specific examples of, like, what
absolutely, like, if you're still doing this byby hand or by yourself, like, you're you're
you're stupid and crazy.
Like, what absolutely should be automated usingAI when when growing your business.
That's you know what?
(21:27):
I will I'm a I'm a bit biased, but I've beenusing, chat JPT for coding for the past 4
months And I I'd swear it saved me at least 20for 20 hours a week just because
it's Wow.
It's so good when it comes to code.
Code is a really good example because it's textand it's logical and structured Andrew so I
certainly save a lot of time.
I also use it now for outlining a lot of myarticles I don't use it for generating content,
(21:51):
but I do gen use it for generating the outlinesor giving the ideas about what to write about.
The reason I don't and I know this isn't reallywhat you asked, but the reason I don't use it
right now is because it still sounds roboticand it still sounds a bit too polished.
Without saying anything.
It says a lot without saying much, but it cando a good job of providing ideas.
But as far as what you should be using it for,coding.
(22:11):
I mean, if there's a software developer outthere who's not using chat to PT today, I would
say you gotta start.
It's really good for idea generation.
So, hey.
I is my business.
Here's a paragraph of context in my business.
Please provide me 20 ideas for what I should bewriting about in terms of, content.
It's really good for that long form outlines.
(22:31):
It's pretty good with that.
Now, what's interesting is you asked me, hey,what should I what do I need to be using it
for?
And I really said a coating Andrew then outsideof that, not much.
And that sort of falls back on, like, this is areally good tool.
But it's not a good job's doer.
And so it depends on what your job is.
You probably have something that you could useAI for but there's nothing that you have to use
(22:54):
it for.
It's just if you use it, it will make you waymore productive and way more efficient probably
save a lot of your own mental effort andenergy.
And so it really just depends on your job.
Yeah.
I I love that.
And, especially because you know, we we thinkthat, we have to be in the perfect mood or the
perfect, room to be able to come up with newideas for what to write about or to talk about
(23:16):
who to talk to, how to talk to them, when AIcan just put you 80 percent of the way there,
you could be like, oh, I know how to fill inthe gaps.
Like, I could definitely customize it formyself.
So That's a really good point you brought up.
And then, I'm curious too.
I know you mostly work with, bigger businesses,but are there any other so the brewers or small
businesses where you where you went in Andrew acase study, you're like, here's what I helped
(23:38):
them with, and here's what also saved them timewith outside of outside of coating.
In the content generation.
Yeah.
I think small businesses what's interesting issmall businesses have been more quick to adopt
this than large because they don't have asmany, lawyers.
They don't have as many risk and and regulatoryissues, and they're really they're really
hungry.
So, for example, one of my first customers samecompany I talked about earlier, government
(24:01):
retirement planning, company called parko, goparko.com.
Check them out.
They're awesome.
What they do, they're a small team, 8 or 9people at most.
And what they do is they meet one on one withevery individual government employee that they
meet with.
And, they've been able to use my tool inconjunction with chat GPT to reduce research
time, I think they said that because they'vegot a tool like this now, they don't need to
(24:23):
hire as many people, and that's because Whenyou've got a tool that plugs into your initial
research documents and can also researchoutside of your industry by using search
engines and chat GPT, you start suddenlycutting down the amount of time that it takes
to get something done.
So that's one example of a small business.
Another example is a small company I work withdown in Rest of Virginia.
(24:44):
They help companies get free money from grants,grants exec.com, go check them out.
And what they do is they help you apply forgrants, and they're using my tool, which is
plugged into their database, to find the bestgrants, grant opportunities for every company
that applies to them.
So what we do is we take your company profileWe we use chat JPT and my tool to match it to
(25:04):
the top 20 or 30 grants that are in theirdatabase of 1000, and then you can find the
best grants to apply for.
So small companies absolutely can use thisstuff.
Even when I said earlier, if you're a smallbusiness owner and you're just doing marketing
or sales, you can use chat with the teamwithout tools like mine.
To just generate the content you need.
So Weiss I suspect big businesses will get alot of efficiency, small businesses today are
(25:25):
the ones who are making the most, progress onthis and really using it to its max.
And it kinda makes sense because if you're asmall business, you have fewer resources, you
have fewer people, the more you can automate,the the better and faster you'll go.
Well, that makes sense.
I love that.
And, and real quick, so tell us more aboutLacoustive and how you can use it with chatgbt
or without chatgbt Andrew that helps companies.
(25:46):
So, I mean, the core problem that we're tryingto solve is when you use chat GPT, it's awesome
for answering things like, hey.
Who wrote this book that was who wroteFrankenstein.
It sucks at saying, how many leads do I have inSalesforce?
Or go look at my product documentation and tellme how I'm different than this customer.
So what Locustin does is we build the datapipelines in the and the software to connect
your data with chat GPU team.
(26:08):
So you can use this to build your own chatbots,or you can use our chatbot and you can connect
these bots to your existing tools likeSalesforce or Google Drive or Google Sheets or
your email, whatever it is.
We'll get all this information We'll read itinto our system, and then you can start asking
questions of it.
So you don't have to go to your Gmail accountAndrew search for that email 30 days ago, or
you don't have to go to your Google Sheets andsay, hey, what row number out of this 1000 row
(26:31):
spreadsheet do I need to look at?
You just ask our bots, or you use our API tobuild your own bots.
And so we provide those infrastructure toolinglayers, and the application layers for
businesses to use large language models withtheir own data.
That as fast meetings.
So could obviously so Google Drive currentlyhas its own search box.
(26:52):
And so you're saying Andrew and, obviously, itis registered by keywords, but if you're
looking for a specific keyword from a specificdocument rather than looking up the keyword,
you know, Jolley or something.
It'll say, well, it's it's here's a ten pageson, with with the word Jolley on it, you're
like, I need to find a Jolley from November 12018 or something like that, and it could,
(27:15):
like, do something like that.
You know, even even more to the point, you cansay, let's take a an example like this in a
real world.
What is the maximum tax rate for retirees thisyear.
Right?
That information is somewhere in your PDFs,Andrew of PDFs in Google Sheets.
All you need to do is ask that specificquestion.
(27:37):
Andrew the bot will come back with the answerand link you to the sheet.
So it's not even you're searching for adocument now.
You're just searching for the thing you need inthat moment.
And the bot will get you that information.
You can still do what you wanted.
Find me the keyword for Jolley, you know, onthis page, on this day, but ultimately what I
found is business owners have a ton of thesethings that they need to look up every day, and
they just want the answer right away.
(27:57):
And so that's what the bot is designed to do.
Now that's so cool.
I could definitely see that saving a lot oftime.
So that's awesome.
Let's talk more about what's going to be mostaffected and least affected by AI.
I know A lot of people are saying, oh, I'mscared of losing my jobs.
You have to rebel against AI now.
And some people are like, oh, I mean, if you'rein you're in sales, if you're in this, like,
(28:21):
it's not gonna affect you kind of thing.
Andrew, tell people, I mean, the ice businessused to be employee sixty thousand people, but
I think it's okay that have freezers now.
Andrew, obviously, I'm sure the horse businesswas great before cars came out and all that.
But based on your knowledge and experience,what it what do you believe is most affected
and affected by AI.
Yeah.
It's such an interesting time right now.
(28:42):
I've heard this as well.
Now I'm I'm actually, optimistic because a lotof folks that I've talked to more in the camp
of, hey.
I'm really excited to use this versus, hey.
I'm worried about losing my job.
So as far as what I what I think is gonnahappen, and I've started to see this a little
bit, I think AI is gonna be able to automate alot of the things that junior employees have
(29:02):
been doing thus far.
So junior software engineers, junior, legal,professionals, accountants.
I think that it's gotten really good.
But it's still at a point where it doesn't do alot of the things that a senior level employee
would do.
So while I am really optimistic, I think thatwhat we might start to see is that companies
start hiring fewer junior level employees andthey give their senior level employees access
(29:26):
to these AI tools and expect them to do more.
I don't know if that's the case.
Because you could also make the argument thatwith these AI tools, you can you can have a
junior level employee who costs a lot Weiss, doa lot more.
So I do suspect we're gonna start to see aredistribution of seniority in companies, and
that goes for a lot of industries.
Right?
You mentioned couple of industries, ice, cars,and, and horses.
(29:51):
I think this is gonna affect a lot ofindustries.
Any sort of thing that's white collar thatrequires information knowledge is going to be
affected by AI.
There are things though that are ultimatelygoing to remain human.
Like, you sew, you buy as a human, and you sellto humans, and so you're going to need people
who are able to connect with humans.
I think that a lot of folks have said, well,writing and copywriting is gonna be affected I
(30:14):
might actually disagree because the writingthat's coming out of these LN's right now is
not good.
It's it's pretty crappy.
And so humans are able to do a lot better of ajob when they come in when it comes to writing
impactful things that are trying to connectwith humans.
Teaching, I think, might be affected in acertain way, but I also think that teachers are
gonna be more empowered to use So I think whatI'm trying to say is every industry is gonna be
(30:36):
impacted, but I suspect what's gonna happen issome jobs might be cut, but a lot of other
jobs, they're gonna benefit from the ability touse AI, and it's one of those things where it
becomes a really good tool for those people whoare working in any industry.
Yeah.
That that that is interesting.
And, for those listening in, we we are talking,it's August 2023 and the rider strike are going
(30:58):
on right now.
Andrew, I think most of the writers are pissedoff that studio executives can, can think, oh,
I can just have chat GPT, write my movies forme, and write my books and and scripts and
stuff.
And so writers are are worried about that.
I know it's tricky too because it it's curiousbecause the writers can use the chat, GPT, how
they want to as well.
(31:18):
And, of course, all comes down to how good youare at prompting Andrew, like you said, filling
in the gaps.
Andrew I think it's definitely, you know,curious and scary.
The fact that we're still in the beginningstages, and as we know, AI and technology is
gonna keep getting better and better.
It sounds like you're still confident thatwe're still a few years, maybe even a decade
out from having competent writing skills froman AI machines.
(31:40):
Oh, what thing I've I've learned to do is notforecast too far in the future.
I used to I used to use AI to predict stock andI quickly got humbled by doing that.
Sometimes it works.
Sometimes it doesn't.
Yeah.
It works not bad.
It works until it doesn't.
It's sort of what I learned.
So I can tell you sort of what I'm seeing rightnow is that long form content is not good from
(32:00):
these AI tools, and it's it's just not good.
Now how far away are we from being able to havereally high quality scripts or plays or
whatever it is.
You know, I don't know.
I don't know.
I think it might be a few years out.
But things are moving so quickly.
It could be a lot less.
I wouldn't I think it will happen at somepoint, but I all so know the way that these
(32:21):
things work under the hood, and I know thatthey're not really optimized for a lot of this
stuff for for, like, long form creativecontent.
Is very difficult to do, especially withcontext windows that are just a few thousand
words or tokens.
So it's a it's a way of me having an on, but Idon't I don't think it's gonna come anytime
soon, and that's where I'll leave it.
(32:41):
No.
That that makes a lot
of sense.
Like, obviously, if you need it to generate aheadline or a paragraph, for some, like, hooks
to something.
Like, it's good at that, but it makes sense.
Like, technology wise, it would take a lot ofwork Andrew data or process putting together a
novel that could be a best seller because it,like you said, it makes sense.
Technology wise, it's very far out.
(33:03):
Is that is that essentially what you're saying?
I think so.
There there are ways to get around this rightnow that some people are trying.
I just talked to somebody who's his brother waswriting a play using chat.
Okay.
In Spanish.
I was like, Weiss, that's kinda cool.
Mhmm.
Now I was thinking, like, how is he doing this?
I think what you have to do is every time youstart a new chapter or Andrew I don't know how
plays are structured.
You have to summarize the old play so that itdoesn't get overloaded, and then you have to
(33:26):
provide a script or an outline for what youwanted to do next.
So maybe there are ways to get around this, buteven then I can't imagine the content to be
that great seeing what I've seen today.
But so, yeah, it's I think the technology isstill a little bit has a ways to go a little
bit.
Well, and and I like that too.
Just like going back to life why you shouldleverage chat dbt is, like, what would take,
(33:46):
you know, 5 years to write a play.
Now you can do it in 1.
Hopefully, and then being able to put out 10plays in 10 years, there's still top quality
versus, trying to figure out every singledetail all the time, essentially.
So that sounds like that's what your is thisyour the first play that your colleague is put
together chat GBT?
You know, this is someone I talked to.
I just got married a couple months ago, andthis is Oh, congrats.
(34:08):
Our photographers, brother, Thank you.
So our photographer's brother in Mexico waswriting a play.
Now I have no idea how that's going.
But I do I do agree with what you said, Andrew,where you can use this in ways to make yourself
way more efficient, and I think that's that'sthe thing I'm harping on is like, how do you
build business tools to do this?
The one thing we haven't really mentioned yetis these tools are actually really good at
(34:32):
making intelligent automated logical trains ofthought, which is something machines have never
been able to do before.
You can have a machine taken a bunch ofinformation, give it a question, and it'll
reason through to get a right answer and thatallows us that opens up a world of
possibilities when it comes to better automatedsoftware.
So I think that you can start to become waymore efficient by having it do your thinking
(34:53):
for you.
Addition to doing your content generation.
No.
I love that.
No.
It makes makes a lot of sense.
Alright.
So let's talk about, yeah.
So you kept bringing up the word tools.
Let's talk about your tools arsenal.
I know people love love hearing the tool stackand and what you use.
So, give us a rundown of you're making yourlife easier with the various tools you're
(35:14):
using.
That's that's interesting.
I use a 20.
I'm a guy who's starting a business on doingeverything from sales to marketing to sending
letters to filing legal docs.
So the main tool I use, at least when I'mcoding, is this tool called, IntelliJ.
It's a company that it's from this company,Jeff Weiss.
I don't know how many of your listeners arecoders, so I don't wanna bore them.
(35:35):
But there isn't a cool tool that plugs into itcalled Copilot, which is AI generated code.
Pretty useful.
I also use chat JBT quite a bit.
So, like, if I have a a bug in my code, I'll gofix I'll go tell Chachi PT to fix it.
That saves me at least 10 minutes or 20 minutesevery time there's a bug because I don't have
to think about As far as everything else, I'musing Notion quite a bit.
You probably hear a lot of people tap thebenefits of Notion.
(35:57):
I run the whole company off of it.
It's our internal Wiki.
And I'm building a Notion integration for mycompany for Locustive so that I don't have to
keep searching through Notion, which is kind offun.
Google Workspace, I use quite a bit Slack.
I use a lot.
I mean, we built our our first bot on on Slackas a chatbot, so use it quite a bit.
Webflow is great for people who need to createwebsites very easy to create websites.
(36:22):
HubSpot, I've started to use a little bit.
There's probably, like, a stack of 10 or 15tools that I wanna bore people that most
startups are using today, and I'm probablyusing about 80 to 90% of those tools.
Have you heard of a gamma app yet that createswebsites for you?
I have not.
No.
I'd like to Yeah.
It's a
(36:42):
it's gamma dot app, and, it it's a tech towebsite tool, which is kind of unique together.
But I I think there's still, I always haven't Ihaven't mastered it yet, but it it is cool.
And I also like these or, presentations too,where you can literally say, make me a
presentation about, pitching my coding companyto, Microsoft or something like that.
(37:03):
And, it'll be able to put together all theslides, all the text, all the images for you.
So, yeah, I haven't played with the web flowyet, but that sounds cool too.
Well, let's
hear about your mentors
Andrew operations that you're learning fromAndrew how'd you come across them?
I think very sort of, randomly, butfortuitously.
(37:26):
So, like, every If you've if you ever hearsomeone say, hey.
I did this on my own, they're lying.
Like, no man is nobody's idling.
Right?
So I think you get people across your careerwho are helping you out in a in a given time.
A season or through your career.
I remember when I first started, there was thisguy I I really looked up to his name as Dan
Usher.
When I worked at Booz Allen, he was really,really, really good when it came to
(37:48):
infrastructure.
So, you know, I I leaned on him quite a bit.
He had already been working at the company.
He took a, hopefully, a light activity, and sothat was sort of my one of my first mentors.
And then, you know, as I started working on TapCommerce, which was that startup, the founders
that I joined up with were really, really,really smart.
So Brian Long, fantastic sales guy, AndrewJones, best AJs, like one of the best product
(38:10):
guys I've ever worked with.
Andrew it just so happened that I had thefortunate good fortune of working with them,
and I learned a lot.
And so for me, it's one of those things whereI've always tried to push myself to, like, try
to get into the next thing that is just justout of reach.
And when you do that, you find people who arealready at that level.
Who can sort of if you if they take a liking toyou or if you work hard enough, they will work
(38:32):
with you to get better.
And so for me, it's always one of those thingswhere I'm always pushing myself, which allows
me to get access to people who are, really goodat what they do and who I can start to learn
from.
So from my perspective, it's one of thosethings where when you work yourself hard
enough, you will eventually come across peoplewho will help you work harder.
Well, I love that.
(38:52):
And and and that for those, listening into,what would you tell them when it comes to
finding a mentor or finding, that, like,community of people who can help support you
and lift you up?
Like, how would you go?
How'd you go about that?
First of all, I think you have to knowyourself.
You know, you have to know, hey, look, this iswhat I care about.
This is what drives me, and it sounds so easy,but It's funny because that's actually the
(39:13):
blocker for a lot of people, particularly intheir twenties.
You don't really know what you wanna do, butlet's say you know what you wanna do.
Let's say you're maybe like me, you're asoftware developer who's always wanted to get
into the startup world.
You can easily find folks like that, and youcan find them on LinkedIn.
You can find them in local meet You can findthem online.
You can find them on other lots of areas.
And so it's it's likely they're not gonna be inyour network.
(39:35):
So you're probably gonna have to do a littlebit of of searching and saying, well, what sort
of companies do I really admire?
What sort of founders do I really care about?
Do I know anybody in my network who might beconnected to start up life?
DCs or investors.
And you kinda have to make a short list offolks who you wanna get connected with, and
then reach out to them and see how you canhelp.
(39:56):
One of the things I've learned in my life isthe more you add value without expecting
anything in return, the faster you're gonnafind people who are really good and can help
you advance your career.
So let's say you find somebody, let's say astartup founder, go offer to run their social
media or go offer to build them free POC or gooffer to do something that you're good at for
free and say, hey.
(40:16):
Look.
I just wanted to get connected.
I wanted to be in your network, and I wanted tosort of show you what I could do.
But the the key is identify what you careabout.
Don't be afraid to reach out in an offer valueoffer value indefinitely and with an unlimited
amount of gusto, and and you'll get a lot ofgood things come back to you.
Well, I I love that, and I love and I one ofthe things I've learned, over the years too is,
(40:41):
like, when you ask people, you know, what whatcan I help with?
What do you need help with?
Like, they don't always wanna put the decisionpower to that.
They're like, oh, I don't know.
I'm good blah blah.
But if you say, hey.
You know, I noticed that your social mediahasn't gotten any posts in the last 2 months,
let me create your next 10 posts for you andsend them over to you.
People are more likely to say yes to that andagreeing to help that.
(41:02):
And so I like what you said too about, like,helping people with something specific rather
than just, like, messaging people and say, whatcan I help with kind of thing?
Yeah.
Yeah.
Because it shows you're authentic and you care.
Right?
Like, it's really easy to say, hey.
What do you need help with?
Then the person who you're asking has to thinkAndrew they have to put mental effort into it.
It's, like, kind of unsure.
If you say, hey.
I'm really good at this.
Can I do this for you then?
(41:22):
Way easier for both of you.
Oh, I love that.
Let's talk about, or Syndrome Andrew.
So I know you mentioned that, you know, doingcode saves you 20 hours a week.
But do you ever present anything to any of yourclients, using AI and be like, oh, is it am I
(41:43):
really this good, or is it just AI making melook good?
Curious if you could talk about this a littlebit.
Yeah, dude.
It's it's funny because like AI is fascinatingto me.
The fact that you can take a bunch of data andpredict, for example, who's gonna churn or
who's gonna buy this product, or what's thenext best word to spit out in the sequence of
words just kinda blows my mind.
(42:04):
So there's almost always a sense of impostorsyndrome.
Like, oh, man, I hope this works this time.
Of course, it's gonna work this time becauseyou've tested it a hundred times.
Like, every time I demo my bots for Slack, andI say, well, you know, let's look let's ask a
question about this piece of information inthis this sheet or this document like thinking,
man, I hope this works this time.
So every time I do this, I'm like, there's acertain sense of fascination that it will work
(42:28):
reliably, and then there's a certain sense of Ireally hope I'm not coming across as a fake.
I really hope I'm being sort of genuine becauseI care about that.
I'm maintaining my integrity.
Because if you're trying to sell, yeah, I'm notthe type of person who can sell anything, so I
still ask them, oh, I'm like, this is what I'vegot.
I hope that this fits your needs.
Hopefully, it does.
So maybe other folks won't have problems withthat, but every time I sort of demo what I'm
(42:52):
doing and I talk about what I'm doing, it isone of those things where it's like, a little
bit too much hesitation, and that's somethingI'm trying to work at, but it's a long way with
me.
I don't wanna ramble.
It's like, yes, imposter syndrome is there.
Even though I've I've been working in datascience for 10 or 15 years.
There's folks out there who are way, way, waymore knowledgeable to me.
You, as a person, just have to stop comparingyourself to those folks.
You just have to say, this is what I could do.
(43:14):
Here's what I've done.
I you know, let's work together.
I promise it'll be a good experience.
And, I like what you said too, that you have toremind yourself, oh, yeah.
I have done this a hundred times.
I have I do know a little bit what I'm what I'mtalking about, and my track record can show it.
And like you said, too, you know, it's alwaysfunny and scary too, taking on new endeavors,
new projects, and and but, you just have tokeep going with what you know, but also trust
(43:36):
yourself to adapt accordingly.
So I like the way you phrased that.
And then I know we have a few minutes lefthere, so I have a few questions to wrap up.
What's the next end game that you're workingon?
It's funny.
I started last year was the was, a year that Ihad to shut down a startup that I had been
working 5 years on.
(43:56):
Andrew if if you had asked me at that point,hey.
Do you wanna do something else?
I'd be like, no way, man.
I am done.
And then, like, for most, like, mostentrepreneurs, I was like, after a few months,
I I said, well, I kinda have this idea.
I wanna see if this works.
I have this thing I wanna work on.
I realized I'm still I still have the bug tobuild something big.
So Locustiv is my attempt at that, but I'mdoing it very differently.
(44:17):
I'm doing it very, very systematically now.
So for me, the the big thing I'm trying to dois turn a lot piece of into a very, very large
company.
And I'm doing it in a way that is not hasn'tbeen done very often before, which sort of
taking it slow, not taking a lot of money untilI really have something that works, moving
very, very systematically.
So, that's what I'm trying to do now.
(44:39):
What what does a a large company mean or looklike?
What's the what's the vision for how how faryou see locus of going?
They're they're sort of the ideal.
And but then there's like, okay.
If that doesn't happen, what happens?
So, like, the ideal is go public and, you know,make it at least, multibillion dollar revenue
company, but that's something that maybe I willrun out of steam.
(45:01):
Maybe I won't have to do that.
Maybe 10 years go by, and I'm at a point whereI don't need that.
For me, a large company is something thatallows me to either draw quite a bit of money
from it on a year year to year basis or sellit.
I don't know what that number is, and Iactually don't wanna put a number on it.
You know?
Whether that's a 100,000,000 or a 1,000,000,000or even 10,000,000 or even 1,000,000.
(45:22):
I'll know it when I see it is the kind of thingthat I've that I'm saying.
But that's sort of what I'm trying to do now isbuild something sustainable and high value
Andrew just enjoy the ride doing it, but alsotry to take it as far as I can.
Wow.
I I love that, and that's, that's definitelyexciting, you know, bringing people in the show
with such big visions and aspirations and wannabuild a $1,000,000,000 company.
(45:45):
And Andrew going into the the mindset ofthings, when Twitter bought your company, could
you have quote unquote retired if you if youwanted to?
Yeah.
You know, I probably don't need to be workingnow, but I am also the type of person.
If I'm not working, I'm like, what am I doing?
Like, I've traveled a lot, which is great.
I've done a lot, but there's a certain sense ofjoy Andrew fulfillment I get from working and
(46:06):
building.
Specifically, I'm a builder.
It just happens that I build with tech andtools and software.
I get a lot of joy from building, and I get alot of joy from seeing people use the things I
build.
And so for me, it's one of those things where Ijust wanna keep pushing myself and seeing how
far I can go.
I love but but what, what helps you feel thatthat hunger to keep working and get that
fulfillment?
(46:27):
I mean, most people are like, oh, yeah.
I'll take the money and keep on chilling.
Like, what what what helps fuel you to staymotivated and keep going?
You know, I bet you I I I wish my wife knew.
I wish I knew.
I Weiss, like, you know, if she knew she wouldbe like, okay.
We can do this without you having to do it thisway.
So because I I'm still working.
I keep track of the time I spend now.
I it's not like 80 hours last week on thisstuff.
(46:47):
So I do some 13 or 14 hour days.
So I don't know.
I just know that that's who I am, and so I usedto fight it Andrew I used to deny it.
And then I was like, well, let's just embraceit.
So I kind of have a sense for why I am the typeperson who wants to keep pushing and keep
getting better, and I find a lot of joy frombeing able to level up.
It's like I played a lot of video games as akid.
(47:10):
But that's just innate to who I am.
And I kind of I've got some ideas of Weiss, butI tried not to dive too deep into that.
At this point.
That concludes another episode of rapidresults.
Remember to leave a review about something youlearn so others can share the keep being
unstoppable in your pursuit of the lifestylefreedom you desire, and we'll see you next
(47:34):
week.
The show you need to get what you desire byavoiding the mistakes made by others before you
learned the stories and journeys of whatsuccess looks like to find a freedom you to
serve Weiss thriving with your best zotes withAndrew Weiss.
(00:32):
I'm back to another episode of rep results withAndrew Weiss.
We have the incredible Shanif Danani here.
And in case you're wondering who this Shanifguy is, He is from New York, and he's an AI and
chat GPT consultant with the background insoftware development and data science.
He helps businesses understand and buildsystems that connect their internal and
proprietary systems to chat DBT, enablinginternal employees, customers, and other
(00:56):
stakeholders to reap the benefits of usinglarge language models on their data.
He's worked with companies like Twitter, BoozAllen, and Tap Commerce, and he now has a CEO
and founder of his own company calledLacoustive.
Did I say that right?
Locustiv.
We got it.
Locustiv.
And so welcome to the show show, Shaniv, tellus what's the biggest and best business deal
(01:17):
you're most proud of.
Oh, man.
Starting right off the bat.
Well, Andrew, thanks for having me.
First of all, it's been an awesome, awesome tobe here.
I apologize for those of you.
I'm in New York.
He might hear sirens in the background.
Oh, yeah.
Very standard for those of us here.
But, no, the the you know, what?
The biggest and coolest thing I've done, we itwas back when we started Tap Commerce, which is
maybe more than 10 years ago at this point.
(01:38):
My background is a tech and AI like you heard,and we built 1 of the world's first mobile
advertising forms at our peak, we were handlingabout a 1,000,000 ad requests a second.
I built a lot of the AI that was using, datascience to figure out how to price these ads as
well as how to respond to them at such a largescale.
So that was really cool.
Kind of the highlight of my tech career.
We eventually sold that company to Twitter, soit all sort of worked out well.
(01:59):
But that's probably the thing I'm really proudabout in terms of something I was able to
accomplish, in my career.
What does that mean a 1,000,000 ad requests asecond.
Oh, it means it means we had a lot of,hardware.
So essentially, you know, the world ofadvertising, you probably have had a a few
guests been in this world.
It is crazy how complicated it is.
(02:21):
When when mobile phones first came out, theystarted to be the this ad ecosystem where
publishers who created apps wanted to get paidby showing ads.
And so there's this ecosystem now where thosepublishers can send an ad request something
like an exchange.
The exchange then fans that request out to abunch of advertiser representatives.
The representatives then have to decide do theywanna show an ad on this product this happens,
(02:45):
you know, this is happening 1000000 of times asecond, and you have to be able to respond in
less than a 100 milliseconds, which is like,how fast you blink your probably faster.
And so it means that we had built up a lot ofsoftware and a lot of infrastructure to be able
to show an ad to the right person at the righttime for the right price.
Oh, that is so cool.
And you know, selling that company to Twitter,what was was that bittersweet, or were you,
(03:08):
like, Victoria C when they sold theircompanies.
Like, jeez, I could've taken this to the moon.
Weiss they have to buy me out like that?
You know, from a from a techie's perspective,it was kinda cool.
It was an opportunity for us see how a largercompany was doing ads at the time.
So he got to work on Twitter's ad platform.
From the perspective of, you know, I'm astartup guy at heart.
So from the perspective of going from somethingsmall to something much, much larger.
(03:33):
I I realized that the larger thing wasn't forme, which is why I went back the startup world,
but it was one of those things where we worked.
That was one of the hardest times I've worked,man.
We were working long hours, 14, 15 hour days,Andrew when you're dealing with a 1,000,000 ad
requests, a lot of your stuff crashes a lot.
And so we had to pick stuff in the middle ofthe night.
I would sometimes get get a buzzer at 3 AM, andI'd have to wake up and start stuff over again.
(03:55):
And so for me, it was a nice respite.
Good transition, but now I'm sort of back at itin the world of startups.
Oh, that that's so cool.
And, I'm very excited to talk about, chat GPTand NII with you today.
So tell us more about this overnight success.
I mean, chat GPT had, what what was it, like, a100,000,000 users in a month or something?
(04:18):
And the next closest competitor.
It was a 100,000,000 users in 2 years.
So what led us to the evolution of how chat tbtbecame so popular in the 1st place?
You know, I don't know, if you guys experiencedthis, but the first time I tried chattyBT, it
was like an moment because I'm a software guy,and a lot of my time is spent looking up stuff
(04:39):
online or trying to fix really nuanced bugs.
It take a lot of time.
And the first time I've tried soft chat JBT tofix the software.
It got me the right answer within a couple ofseconds.
And so I think that's emblematic of why thisthing has blown up.
You know, you're suddenly getting huge amountsof value.
You're saving lots of time.
Andrew you're getting a lot of actually reallygood information at your fingertips in seconds,
(04:59):
whereas it might have taken even 4 or 5 hoursto find a piece of information before,
certainly the case in software development.
Andrew so, you know, they say that everyovernight success is 10 years in the making.
Right?
So GPT 3.5, which came out earlier this year,was following 3 that followed 2, and they've
actually been working on this stuff for a longtime, lots of years.
But, 3.5 was so good that I think it just got alot of people really excited and really
(05:23):
interested, and it just sort of the hype blewup.
And it it it was really cool to see and cool tobe part of.
At least it's interesting for me to be part ofthat process now.
That's so cool.
And how did you get involved in the space?
Obviously, you know, you have a lot ofexperience in tech and e commerce and all
these, language learning models.
But how did you how were you able to dive inand say, oh, I'm a consultant for this because
(05:46):
I know exactly how to use it.
Yeah.
You know, what's funny is, I started using itAndrew then I found that it didn't do a few
things that I wanted.
I like for example, I couldn't just ask it,hey, how many leads do I have in Salesforce, or
how many what's in my product documentation,which lives on Google Drive, give me the
contents of page 33.
So like what I started seeing was a lot ofbusiness owners wanted to use this, but they
(06:10):
couldn't because a lot of their docs werebehind firewall or you needed to log in.
And by default, chatTPT doesn't have access tothese things.
And so it was kinda cool for me because mybackground is in AI and software and data.
So I was able to build the pipes that connectedyour Google Drive or your Salesforce account to
chat GPT.
And a lot of people were like, oh, that's kindacool.
Now can I do this for my, you know, GoogleSheets Andrew can I do this for my email Andrew
(06:34):
And can you add this from my Slack workspace asa chatbot?
And so a lot of folks were just starting to getinterested in what became possible by
connecting their data to chat GT Andrew becauseI had I have a lot of experience building those
sorts of data tools, I was able to jump in, andhopefully deliver something that they
appreciate.
I love that.
And that's the best way to, yeah, get good atthings is to see where the gaps are and go.
(06:58):
Well, I can fix that.
I know how to I know I know how to, adjust forthat.
So so let's talk more about to so you yourjourney with chat TBT is, you're like, oh, this
is so cool.
It can do something in seconds, which I've beenin a career would usually take hours.
It's still missing things.
I can figure out how to close those gaps.
So so tell us more about, what what do you wishyou knew earlier when you're just starting off
(07:23):
with the using chat g b t in the first place.
You know, it's funny that they're asking thatbecause we're still so early in the world of
chat g b t.
But 6 months almost feels like 10 years, right,from a few like, things move so quickly that
now a short amount of time feels like it wasthe result of a very large, large amount of
sense.
So for me, what do I wish I knew earlier thisyear?
(07:43):
I think that there's, there's still a lot ofThere's still a lot of things that's that I
struggle with when it comes to implementingchat TPT on my own data.
For example, one of the biggest limit right nowis that you can only send in a few thousand
words every time you wanna get an answer, andyou also, whenever you ask an answer, you also
need provide the right context for chat dbt toanswer you using your data.
(08:08):
I passed, with with
the chat dbt update.
I was told you could upload, like, a book, thesize of the lord of the rings, or is that not
come come through yet?
That's what they say, man, but it's 32,000tokens equates to about 8000 words, which is
maybe 20 or 30 pages, maybe 40 pages.
But Oh.
You know, essentially, you're really limited.
Right?
So the this would be awesome if I could.
(08:28):
If I could just take all one thousand pages ofmy product handbook and dump it into chat, you
could see, and ask the question about somethingthat lives on page 534.
We're good, but, but you can't.
And that's a lot of the things that businessesare running into now.
So, you know, I wish I had known that before,but in a certain sense, it's It's also provided
me with an opportunity to build the softwarethat lets people find the right, pieces of
(08:50):
their data that I can send to chat GPT so thatthey can actually ask a question.
So that's one thing.
Another thing is, hey, chat t p t if you use itthrough the API, they don't let you send a lot
of information.
They don't let you send a lot of requests.
You kind of get what's called rate limited ortimed out a lot, which means you have to sort
of take your time when you send them stuffAndrew it makes your product, my product, a
(09:11):
little bit slower for users.
Still usable by by far, but still one of thosethings I wish I had known earlier.
And there's all sorts of things like this.
Now I'm nitpicking.
Right?
This is such a cool technology.
It's all interesting.
These are really minor things that I'm sureOpenAI is gonna fix.
But the point is if you're trying to build ontop chat JPT or you're trying to use it for
things like connecting your own data, you haveto be ready for a lot of these challenges.
(09:35):
And if you're not, maybe it's better to use athird party tool.
No.
That that makes it makes a lot of sense.
So just being aware of what it can and can'tdo, it's definitely definitely good to know.
And I know you bring yourself as a chat tint,an LOM consultant, but how much have you played
around with, Bard Andrew Claude and and allthose?
I'm curious your your quick thoughts on allthose.
(09:56):
I like LAMA.
It's super fast.
I love Claude.
You know, Claude is one of those things forthose who don't know.
It's from this company called Antropic,competitor to chat JPT.
Claud actually lets you put in something like75,000 token, which is, 2 to 3 times what, you
know, what chat TPT 4 lets you do.
For what it's worth, I use chat TPT 3.5 withthe 16,000 token limit because it's much
(10:17):
cheaper.
I digress.
Cloud's awesome.
The reason it's awesome is because because youcan put in a lot more, information, and it's
super fast.
You can get sort of a lot of long form contentfrom it a little bit more quickly and a little
bit more accurately.
These LOM still aren't great for long formcontent.
They're still a little bit robotic and generic,but cloud does a really good job of providing
(10:39):
you with, like, logs or articles ornewsletters, you'll still have to edit them,
but it's a good start.
Llama is really fast.
Llama's a the open source tool from Facebook,really good Bard, I've used a little bit.
I didn't think Bard sort of got a little wonkyfor me when I tried it, but this was about a
month or 2 ago.
So I'm sure it's gotten a lot better now.
Shows you how fast things are going.
I think the whole ecosystem, the fact thatthere are so many of these providers is
(11:02):
awesome, particularly for companies that are inregulated fields like medicine because they're
gonna need to build their own open sourceversions of this or at least implement them
because they can't send data outside theirsystems.
So it's a long way of me saying, I think thesethese tools are all really cool.
What what I got out of that is it sounds likeif you have a lot of, words and, like, product
information, you have to pull from or summarizeor stuff like that, cause the way to go because
(11:28):
you can upload a lot more tokens to get moreinformation.
It sounds like chat, GPT, though.
They're still top for, like, create activitywriting and, like, being able to, like, match
your writing style per se.
Andrew sounds like llama's kind of a mixture ofthose.
Is that kind of a 3rd grade summary of things.
It's it's a good, like, yeah, it's a good highlevel approach.
You know, obviously, there are nuances, detailswith everything.
(11:50):
For so for example, a lot of people don't thinkthought is as good with software development.
So I I I probably would fall in that camp, butI still use chat JPT for a lot of my coding.
I use Claude for a lot of my long form content.
Llama, I use every now and then mostly for thenovelty.
Bard, I don't use as much unfortunately, but II'm sure I will as Google makes it better.
(12:11):
And so it's one of those things where you sortof get used to doing different things with
different tools, and you're probably gonnastick with that until then, unless another tool
comes out, that's just a lot better.
One thing to keep in mind, though, is evenClaude, which can take a lot more content, it's
still limited, and it's still limited to maybe75,000 tokens, which is about, I don't know,
20,000 words or so to give or take, 13,000.
(12:33):
You're still not gonna be able to dump a wholelot of information into it.
And so that's one of the biggest issues withbusiness today is if you've got a lot of
documentation, How can you make sure that youcan answer questions from that documentation
given these what's called context windows ortoken limits?
Interesting.
And you Andrew you keep saying that tokens, youknow, why does it have to be tokens versus
(12:53):
just, like, words or characters, essentially?
Yeah.
It's, I apologize.
When you're sort of in this field, you startusing the jargon, these tool, these companies
like OpenAI Andrew Google and and propic,they've they've introduced this concept of a
token, which is Kind of like a couple ofletters or a part of a word or an individual
(13:15):
character like maybe a semi colon.
Now the reason they use tokens instead of fullwords is because the AI that runs this stuff
basically processes the text that you get insmall chunks and what they they call these
things tokens Andrew when you have a token,it's actually a lot easier to turn those tokens
into some sort of numerical representation thatcomputers are better at dealing with than if
(13:37):
you had a whole word.
So you might hear the word token in general,one token.
They say 4 tokens is maybe a word.
I'm sorry.
They say 4 characters is about one token.
And so that's how you can start to think aboutthis stuff.
Interesting.
Okay.
That's good to know.
So let's talk about, you know, I know a lot ofpeople are still, you know, even though with a
(13:59):
100,000,000 Andrew counting and maybe it's at200,000,000 by now of users of AI, there's
still a lot of talk in the industry of Like, Idon't know if I can use AI.
It sounds too complicated for me.
What are the all the ethical implications withit?
So what do you tell people to help comfort themabout being okay with using AI and leveraging
it?
You know, the first thing I tell them is if youdon't do it, somebody else will.
(14:21):
So, you know, you're you're probably, I mean,unless you're close to retired or you don't
care anymore, then it's fine.
Right?
Who cares?
But if you're really, you know, don't know andyou're you're very career oriented, like, like
those of us in New York tend to be, youprobably wanna get the best tools for your job.
And AI is probably gonna be in that toolkit atsome point if it's not already.
(14:43):
So I would generally say I understand thatthere's some sort of hesitancy to using it, but
also you can just go and start using it rightaway, and it's very easy, and it actually is
very simple to use.
And so that's how I started.
I just went over to chat TPT's website.
They give you a few credits, you know, for freeto get started, and I just started asking it a
few questions.
(15:04):
Andrew once when I had a coding issue, I hadsaid, hey.
How do I solve this?
And it it came back with the right answer.
And then I started using this tool calledCopilot, which is a after code.
And so, like, you can start really simply.
You don't need to dive all in.
The folks who make these tools make it reallyeasy to get started, and you're only doing
yourself a disservice if you don't know whatthey are.
And so it's it really behooms you to kinda getfamiliar with these tools.
(15:26):
Now maybe you say, hey.
Look.
I tried it.
It has no bearing on my work.
It doesn't make any sense for me.
There are folks like that.
That's fine.
At least you've tried it.
I would I would argue that's probably not thecase for most people, but sure if that's your
opinion, sure.
But you gotta at least try it because it'seasy.
It's free.
It takes 2 seconds, and it probably wouldchange your life at some point.
Yeah.
And I love, the analogy of, know, sometimes,you know, AI only puts out what you put into it
(15:53):
essentially.
And so if if you, like, tell it, you know, makeme a $1,000,000.
You know, it's probably gonna be like, what?
But it, obviously, it's to keep it.
I think some person said it's kinda a atoddler, like, hey.
Can you, go collect this book from thisbookshelf and bring this Sunrise information,
or can you solve this little problem for me?
It's like, okay.
Yeah.
I got you kinda thing.
(16:14):
Andrew was following a tweet too about, like,the hustle GPT guide.
Did you hear about that at all?
No.
I haven't
so he basically was doing he started a Twittertrend.
He's like, okay.
I'm gonna see if Chad GPT can help me make a$1,000,000.
So I'm gonna name it, hustled GPT Andrew tellme what to do every day for 30 days straight.
And he essentially made all of his money offpeople joining his community.
And so technically, you you he's getting closerand closer.
(16:41):
But but just, yeah, it's it's a reminder toothat, like, you know, it it can only put out
what we put into it Andrew just, like,constantly having someone like Shanif to help
you, like, tweak it and make sure you get thebest out of it.
Sure, is definitely essential.
Andrew so speaking of that, yeah, she tellssome more examples.
Let's say you're you're a business owner andyou're like, alright.
How can I be better at marketing my business?
(17:03):
Andrew I sell, let's just say you you sellvoice, coaching lessons, essentially.
How how would you use a chat GPT to build yourvoice, voice coaching business?
You know, this is it's a really interestingquestion.
Because a lot of businesses want to use thistool, but they don't know how.
And I almost always recommend them to startwith a small problem that they have So take a
(17:27):
look at what you do on a on a day to day basis.
And if there's something that has just taken upa little bit of time, 20, 30 minutes at a time,
but you have to do it a lot.
Andrew maybe you might be able to automate it,but you probably have to do it yourself because
it requires a little bit of thought.
That's a really good opportunity to use chatJPT.
Now maybe that's not necessarily growing ormarketing your business.
Maybe that's operations, but I bet you if youfind something that you're doing a lot, that's
(17:51):
taking up a little bit of like, a good amountof time, and it's not really necessary.
It's a bit frustrating.
You probably could automate that away usingsomething like chat GPT.
Now, if you really wanted to start creating newrevenue generating opportunities, there's a few
ways you could apply this, like say, let's justbreak it down sales, marketing, and product.
On the sales side, you've got folks today whoare using chat JPT to help you write emails.
(18:16):
Lot of times, it's gonna be personalizedemails.
Like, you can copy and paste somebody'sLinkedIn profile and say, hey, chat to PT,
like, you know, create a personalized email forthis.
A lot of times, you can start doingpersonalization at scale.
There's a few tools out there right now thatlet you scrape someone's LinkedIn and then
write an email and do this several 100 or 1000times a day.
What
tools are those?
That sounds pretty valuable.
(18:37):
There's I think there's one I started lookingat called reply.i0, if I remember correctly.
I haven't used this myself.
However, it's something, yeah, it's somethingthat I wanna use at some point soon, drive your
revenue growth with Weiss So replying.i0 willlet you automatically personalize things at
scale.
As far as I know, there's a few other tools outthere, that lets you do something like this
(18:58):
similar.
I have no affiliation with the Play Datta.
I just saw them Monday.
And then on the mark, let's see.
On another way that sales folks at least arestarting to use my stuff, is a lot of times
they'll get these really obscure questions fromtheir customers on a call.
For example, like one of my customers doesfinancial retirement planning or government
employees, Andrew the government has all sortsof crazy policies.
(19:19):
So, like, the customer, theirs might say, hey.
I'm a I'm a spouse of a surviving military, youknow, benefits, receiver, I I just hit seventy
years old.
Can I start borrowing from my federalretirement account next month?
Like like, what?
Andrew so these guys are on a call.
They get a co a question like this on theircustomer.
What they were doing before was they would haveto message their team internally to look this
(19:42):
up up Andrew then they tell the customer, hey.
I'll get back to you.
Now because they have a tool like mine that canconnect their internal policy documents to a
chatbot, they can just ask the chatbot andthey'll get an answer back in a few seconds,
and they can tell the customer the answer rightaway.
And so sales guys are now using tools like thison their calls to answer customer questions.
They're using it to transcribe their calls andget a sense for what messages work and what
(20:05):
they'll So, again, it you talk about, hey.
How does chat JPT grow my business?
I like to say it's really chat JPT and othertools like this are really good at tasks.
But they suck at jobs.
So if you give them a job, like you said, tella toddler to go get my book from the shelf and
tell me what's on page 3.
It's gonna be great.
But if you say, like, you said, hey.
How do I make me a $1,000,000?
(20:26):
It's not gonna come back and tell you what todo, or maybe it is if you're this one guy.
But if you're anyone else, it's not gonna tellyou anything.
So you as the business owner need to know whatyour job is, what are your tasks in that job,
where can you start to automate some of thisstuff away, and how can you use automation
right now to start growing in those areas.
Yeah.
It's a long way of answering your question.
No.
(20:46):
And and it's great because you you mentioned,well, how how does operations grow your
business?
Mean, I would say, well, if you're able toautomate your operations, that gives you more
time for marketing, more time for fulfillment,more time for sales, and be able to keep
outsourcing time to other things that stillbecause it still takes time to grow your
business.
If you're so a brand new or small business, yougotta wear many hats where it's always better
(21:07):
to outsource and delegate those other hats.
So my next question was asking, can Andrew givemore, specific examples of, like, what
absolutely, like, if you're still doing this byby hand or by yourself, like, you're you're
you're stupid and crazy.
Like, what absolutely should be automated usingAI when when growing your business.
That's you know what?
(21:27):
I will I'm a I'm a bit biased, but I've beenusing, chat JPT for coding for the past 4
months And I I'd swear it saved me at least 20for 20 hours a week just because
it's Wow.
It's so good when it comes to code.
Code is a really good example because it's textand it's logical and structured Andrew so I
certainly save a lot of time.
I also use it now for outlining a lot of myarticles I don't use it for generating content,
(21:51):
but I do gen use it for generating the outlinesor giving the ideas about what to write about.
The reason I don't and I know this isn't reallywhat you asked, but the reason I don't use it
right now is because it still sounds roboticand it still sounds a bit too polished.
Without saying anything.
It says a lot without saying much, but it cando a good job of providing ideas.
But as far as what you should be using it for,coding.
(22:11):
I mean, if there's a software developer outthere who's not using chat to PT today, I would
say you gotta start.
It's really good for idea generation.
So, hey.
I is my business.
Here's a paragraph of context in my business.
Please provide me 20 ideas for what I should bewriting about in terms of, content.
It's really good for that long form outlines.
(22:31):
It's pretty good with that.
Now, what's interesting is you asked me, hey,what should I what do I need to be using it
for?
And I really said a coating Andrew then outsideof that, not much.
And that sort of falls back on, like, this is areally good tool.
But it's not a good job's doer.
And so it depends on what your job is.
You probably have something that you could useAI for but there's nothing that you have to use
(22:54):
it for.
It's just if you use it, it will make you waymore productive and way more efficient probably
save a lot of your own mental effort andenergy.
And so it really just depends on your job.
Yeah.
I I love that.
And, especially because you know, we we thinkthat, we have to be in the perfect mood or the
perfect, room to be able to come up with newideas for what to write about or to talk about
(23:16):
who to talk to, how to talk to them, when AIcan just put you 80 percent of the way there,
you could be like, oh, I know how to fill inthe gaps.
Like, I could definitely customize it formyself.
So That's a really good point you brought up.
And then, I'm curious too.
I know you mostly work with, bigger businesses,but are there any other so the brewers or small
businesses where you where you went in Andrew acase study, you're like, here's what I helped
(23:38):
them with, and here's what also saved them timewith outside of outside of coating.
In the content generation.
Yeah.
I think small businesses what's interesting issmall businesses have been more quick to adopt
this than large because they don't have asmany, lawyers.
They don't have as many risk and and regulatoryissues, and they're really they're really
hungry.
So, for example, one of my first customers samecompany I talked about earlier, government
(24:01):
retirement planning, company called parko, goparko.com.
Check them out.
They're awesome.
What they do, they're a small team, 8 or 9people at most.
And what they do is they meet one on one withevery individual government employee that they
meet with.
And, they've been able to use my tool inconjunction with chat GPT to reduce research
time, I think they said that because they'vegot a tool like this now, they don't need to
(24:23):
hire as many people, and that's because Whenyou've got a tool that plugs into your initial
research documents and can also researchoutside of your industry by using search
engines and chat GPT, you start suddenlycutting down the amount of time that it takes
to get something done.
So that's one example of a small business.
Another example is a small company I work withdown in Rest of Virginia.
(24:44):
They help companies get free money from grants,grants exec.com, go check them out.
And what they do is they help you apply forgrants, and they're using my tool, which is
plugged into their database, to find the bestgrants, grant opportunities for every company
that applies to them.
So what we do is we take your company profileWe we use chat JPT and my tool to match it to
(25:04):
the top 20 or 30 grants that are in theirdatabase of 1000, and then you can find the
best grants to apply for.
So small companies absolutely can use thisstuff.
Even when I said earlier, if you're a smallbusiness owner and you're just doing marketing
or sales, you can use chat with the teamwithout tools like mine.
To just generate the content you need.
So Weiss I suspect big businesses will get alot of efficiency, small businesses today are
(25:25):
the ones who are making the most, progress onthis and really using it to its max.
And it kinda makes sense because if you're asmall business, you have fewer resources, you
have fewer people, the more you can automate,the the better and faster you'll go.
Well, that makes sense.
I love that.
And, and real quick, so tell us more aboutLacoustive and how you can use it with chatgbt
or without chatgbt Andrew that helps companies.
(25:46):
So, I mean, the core problem that we're tryingto solve is when you use chat GPT, it's awesome
for answering things like, hey.
Who wrote this book that was who wroteFrankenstein.
It sucks at saying, how many leads do I have inSalesforce?
Or go look at my product documentation and tellme how I'm different than this customer.
So what Locustin does is we build the datapipelines in the and the software to connect
your data with chat GPU team.
(26:08):
So you can use this to build your own chatbots,or you can use our chatbot and you can connect
these bots to your existing tools likeSalesforce or Google Drive or Google Sheets or
your email, whatever it is.
We'll get all this information We'll read itinto our system, and then you can start asking
questions of it.
So you don't have to go to your Gmail accountAndrew search for that email 30 days ago, or
you don't have to go to your Google Sheets andsay, hey, what row number out of this 1000 row
(26:31):
spreadsheet do I need to look at?
You just ask our bots, or you use our API tobuild your own bots.
And so we provide those infrastructure toolinglayers, and the application layers for
businesses to use large language models withtheir own data.
That as fast meetings.
So could obviously so Google Drive currentlyhas its own search box.
(26:52):
And so you're saying Andrew and, obviously, itis registered by keywords, but if you're
looking for a specific keyword from a specificdocument rather than looking up the keyword,
you know, Jolley or something.
It'll say, well, it's it's here's a ten pageson, with with the word Jolley on it, you're
like, I need to find a Jolley from November 12018 or something like that, and it could,
(27:15):
like, do something like that.
You know, even even more to the point, you cansay, let's take a an example like this in a
real world.
What is the maximum tax rate for retirees thisyear.
Right?
That information is somewhere in your PDFs,Andrew of PDFs in Google Sheets.
All you need to do is ask that specificquestion.
(27:37):
Andrew the bot will come back with the answerand link you to the sheet.
So it's not even you're searching for adocument now.
You're just searching for the thing you need inthat moment.
And the bot will get you that information.
You can still do what you wanted.
Find me the keyword for Jolley, you know, onthis page, on this day, but ultimately what I
found is business owners have a ton of thesethings that they need to look up every day, and
they just want the answer right away.
(27:57):
And so that's what the bot is designed to do.
Now that's so cool.
I could definitely see that saving a lot oftime.
So that's awesome.
Let's talk more about what's going to be mostaffected and least affected by AI.
I know A lot of people are saying, oh, I'mscared of losing my jobs.
You have to rebel against AI now.
And some people are like, oh, I mean, if you'rein you're in sales, if you're in this, like,
(28:21):
it's not gonna affect you kind of thing.
Andrew, tell people, I mean, the ice businessused to be employee sixty thousand people, but
I think it's okay that have freezers now.
Andrew, obviously, I'm sure the horse businesswas great before cars came out and all that.
But based on your knowledge and experience,what it what do you believe is most affected
and affected by AI.
Yeah.
It's such an interesting time right now.
(28:42):
I've heard this as well.
Now I'm I'm actually, optimistic because a lotof folks that I've talked to more in the camp
of, hey.
I'm really excited to use this versus, hey.
I'm worried about losing my job.
So as far as what I what I think is gonnahappen, and I've started to see this a little
bit, I think AI is gonna be able to automate alot of the things that junior employees have
(29:02):
been doing thus far.
So junior software engineers, junior, legal,professionals, accountants.
I think that it's gotten really good.
But it's still at a point where it doesn't do alot of the things that a senior level employee
would do.
So while I am really optimistic, I think thatwhat we might start to see is that companies
start hiring fewer junior level employees andthey give their senior level employees access
(29:26):
to these AI tools and expect them to do more.
I don't know if that's the case.
Because you could also make the argument thatwith these AI tools, you can you can have a
junior level employee who costs a lot Weiss, doa lot more.
So I do suspect we're gonna start to see aredistribution of seniority in companies, and
that goes for a lot of industries.
Right?
You mentioned couple of industries, ice, cars,and, and horses.
(29:51):
I think this is gonna affect a lot ofindustries.
Any sort of thing that's white collar thatrequires information knowledge is going to be
affected by AI.
There are things though that are ultimatelygoing to remain human.
Like, you sew, you buy as a human, and you sellto humans, and so you're going to need people
who are able to connect with humans.
I think that a lot of folks have said, well,writing and copywriting is gonna be affected I
(30:14):
might actually disagree because the writingthat's coming out of these LN's right now is
not good.
It's it's pretty crappy.
And so humans are able to do a lot better of ajob when they come in when it comes to writing
impactful things that are trying to connectwith humans.
Teaching, I think, might be affected in acertain way, but I also think that teachers are
gonna be more empowered to use So I think whatI'm trying to say is every industry is gonna be
(30:36):
impacted, but I suspect what's gonna happen issome jobs might be cut, but a lot of other
jobs, they're gonna benefit from the ability touse AI, and it's one of those things where it
becomes a really good tool for those people whoare working in any industry.
Yeah.
That that that is interesting.
And, for those listening in, we we are talking,it's August 2023 and the rider strike are going
(30:58):
on right now.
Andrew, I think most of the writers are pissedoff that studio executives can, can think, oh,
I can just have chat GPT, write my movies forme, and write my books and and scripts and
stuff.
And so writers are are worried about that.
I know it's tricky too because it it's curiousbecause the writers can use the chat, GPT, how
they want to as well.
(31:18):
And, of course, all comes down to how good youare at prompting Andrew, like you said, filling
in the gaps.
Andrew I think it's definitely, you know,curious and scary.
The fact that we're still in the beginningstages, and as we know, AI and technology is
gonna keep getting better and better.
It sounds like you're still confident thatwe're still a few years, maybe even a decade
out from having competent writing skills froman AI machines.
(31:40):
Oh, what thing I've I've learned to do is notforecast too far in the future.
I used to I used to use AI to predict stock andI quickly got humbled by doing that.
Sometimes it works.
Sometimes it doesn't.
Yeah.
It works not bad.
It works until it doesn't.
It's sort of what I learned.
So I can tell you sort of what I'm seeing rightnow is that long form content is not good from
(32:00):
these AI tools, and it's it's just not good.
Now how far away are we from being able to havereally high quality scripts or plays or
whatever it is.
You know, I don't know.
I don't know.
I think it might be a few years out.
But things are moving so quickly.
It could be a lot less.
I wouldn't I think it will happen at somepoint, but I all so know the way that these
(32:21):
things work under the hood, and I know thatthey're not really optimized for a lot of this
stuff for for, like, long form creativecontent.
Is very difficult to do, especially withcontext windows that are just a few thousand
words or tokens.
So it's a it's a way of me having an on, but Idon't I don't think it's gonna come anytime
soon, and that's where I'll leave it.
(32:41):
No.
That that makes a lot
of sense.
Like, obviously, if you need it to generate aheadline or a paragraph, for some, like, hooks
to something.
Like, it's good at that, but it makes sense.
Like, technology wise, it would take a lot ofwork Andrew data or process putting together a
novel that could be a best seller because it,like you said, it makes sense.
Technology wise, it's very far out.
(33:03):
Is that is that essentially what you're saying?
I think so.
There there are ways to get around this rightnow that some people are trying.
I just talked to somebody who's his brother waswriting a play using chat.
Okay.
In Spanish.
I was like, Weiss, that's kinda cool.
Mhmm.
Now I was thinking, like, how is he doing this?
I think what you have to do is every time youstart a new chapter or Andrew I don't know how
plays are structured.
You have to summarize the old play so that itdoesn't get overloaded, and then you have to
(33:26):
provide a script or an outline for what youwanted to do next.
So maybe there are ways to get around this, buteven then I can't imagine the content to be
that great seeing what I've seen today.
But so, yeah, it's I think the technology isstill a little bit has a ways to go a little
bit.
Well, and and I like that too.
Just like going back to life why you shouldleverage chat dbt is, like, what would take,
(33:46):
you know, 5 years to write a play.
Now you can do it in 1.
Hopefully, and then being able to put out 10plays in 10 years, there's still top quality
versus, trying to figure out every singledetail all the time, essentially.
So that sounds like that's what your is thisyour the first play that your colleague is put
together chat GBT?
You know, this is someone I talked to.
I just got married a couple months ago, andthis is Oh, congrats.
(34:08):
Our photographers, brother, Thank you.
So our photographer's brother in Mexico waswriting a play.
Now I have no idea how that's going.
But I do I do agree with what you said, Andrew,where you can use this in ways to make yourself
way more efficient, and I think that's that'sthe thing I'm harping on is like, how do you
build business tools to do this?
The one thing we haven't really mentioned yetis these tools are actually really good at
(34:32):
making intelligent automated logical trains ofthought, which is something machines have never
been able to do before.
You can have a machine taken a bunch ofinformation, give it a question, and it'll
reason through to get a right answer and thatallows us that opens up a world of
possibilities when it comes to better automatedsoftware.
So I think that you can start to become waymore efficient by having it do your thinking
(34:53):
for you.
Addition to doing your content generation.
No.
I love that.
No.
It makes makes a lot of sense.
Alright.
So let's talk about, yeah.
So you kept bringing up the word tools.
Let's talk about your tools arsenal.
I know people love love hearing the tool stackand and what you use.
So, give us a rundown of you're making yourlife easier with the various tools you're
(35:14):
using.
That's that's interesting.
I use a 20.
I'm a guy who's starting a business on doingeverything from sales to marketing to sending
letters to filing legal docs.
So the main tool I use, at least when I'mcoding, is this tool called, IntelliJ.
It's a company that it's from this company,Jeff Weiss.
I don't know how many of your listeners arecoders, so I don't wanna bore them.
(35:35):
But there isn't a cool tool that plugs into itcalled Copilot, which is AI generated code.
Pretty useful.
I also use chat JBT quite a bit.
So, like, if I have a a bug in my code, I'll gofix I'll go tell Chachi PT to fix it.
That saves me at least 10 minutes or 20 minutesevery time there's a bug because I don't have
to think about As far as everything else, I'musing Notion quite a bit.
You probably hear a lot of people tap thebenefits of Notion.
(35:57):
I run the whole company off of it.
It's our internal Wiki.
And I'm building a Notion integration for mycompany for Locustive so that I don't have to
keep searching through Notion, which is kind offun.
Google Workspace, I use quite a bit Slack.
I use a lot.
I mean, we built our our first bot on on Slackas a chatbot, so use it quite a bit.
Webflow is great for people who need to createwebsites very easy to create websites.
(36:22):
HubSpot, I've started to use a little bit.
There's probably, like, a stack of 10 or 15tools that I wanna bore people that most
startups are using today, and I'm probablyusing about 80 to 90% of those tools.
Have you heard of a gamma app yet that createswebsites for you?
I have not.
No.
I'd like to Yeah.
It's a
(36:42):
it's gamma dot app, and, it it's a tech towebsite tool, which is kind of unique together.
But I I think there's still, I always haven't Ihaven't mastered it yet, but it it is cool.
And I also like these or, presentations too,where you can literally say, make me a
presentation about, pitching my coding companyto, Microsoft or something like that.
(37:03):
And, it'll be able to put together all theslides, all the text, all the images for you.
So, yeah, I haven't played with the web flowyet, but that sounds cool too.
Well, let's
hear about your mentors
Andrew operations that you're learning fromAndrew how'd you come across them?
I think very sort of, randomly, butfortuitously.
(37:26):
So, like, every If you've if you ever hearsomeone say, hey.
I did this on my own, they're lying.
Like, no man is nobody's idling.
Right?
So I think you get people across your careerwho are helping you out in a in a given time.
A season or through your career.
I remember when I first started, there was thisguy I I really looked up to his name as Dan
Usher.
When I worked at Booz Allen, he was really,really, really good when it came to
(37:48):
infrastructure.
So, you know, I I leaned on him quite a bit.
He had already been working at the company.
He took a, hopefully, a light activity, and sothat was sort of my one of my first mentors.
And then, you know, as I started working on TapCommerce, which was that startup, the founders
that I joined up with were really, really,really smart.
So Brian Long, fantastic sales guy, AndrewJones, best AJs, like one of the best product
(38:10):
guys I've ever worked with.
Andrew it just so happened that I had thefortunate good fortune of working with them,
and I learned a lot.
And so for me, it's one of those things whereI've always tried to push myself to, like, try
to get into the next thing that is just justout of reach.
And when you do that, you find people who arealready at that level.
Who can sort of if you if they take a liking toyou or if you work hard enough, they will work
(38:32):
with you to get better.
And so for me, it's always one of those thingswhere I'm always pushing myself, which allows
me to get access to people who are, really goodat what they do and who I can start to learn
from.
So from my perspective, it's one of thosethings where when you work yourself hard
enough, you will eventually come across peoplewho will help you work harder.
Well, I love that.
(38:52):
And and and that for those, listening into,what would you tell them when it comes to
finding a mentor or finding, that, like,community of people who can help support you
and lift you up?
Like, how would you go?
How'd you go about that?
First of all, I think you have to knowyourself.
You know, you have to know, hey, look, this iswhat I care about.
This is what drives me, and it sounds so easy,but It's funny because that's actually the
(39:13):
blocker for a lot of people, particularly intheir twenties.
You don't really know what you wanna do, butlet's say you know what you wanna do.
Let's say you're maybe like me, you're asoftware developer who's always wanted to get
into the startup world.
You can easily find folks like that, and youcan find them on LinkedIn.
You can find them in local meet You can findthem online.
You can find them on other lots of areas.
And so it's it's likely they're not gonna be inyour network.
(39:35):
So you're probably gonna have to do a littlebit of of searching and saying, well, what sort
of companies do I really admire?
What sort of founders do I really care about?
Do I know anybody in my network who might beconnected to start up life?
DCs or investors.
And you kinda have to make a short list offolks who you wanna get connected with, and
then reach out to them and see how you canhelp.
(39:56):
One of the things I've learned in my life isthe more you add value without expecting
anything in return, the faster you're gonnafind people who are really good and can help
you advance your career.
So let's say you find somebody, let's say astartup founder, go offer to run their social
media or go offer to build them free POC or gooffer to do something that you're good at for
free and say, hey.
(40:16):
Look.
I just wanted to get connected.
I wanted to be in your network, and I wanted tosort of show you what I could do.
But the the key is identify what you careabout.
Don't be afraid to reach out in an offer valueoffer value indefinitely and with an unlimited
amount of gusto, and and you'll get a lot ofgood things come back to you.
Well, I I love that, and I love and I one ofthe things I've learned, over the years too is,
(40:41):
like, when you ask people, you know, what whatcan I help with?
What do you need help with?
Like, they don't always wanna put the decisionpower to that.
They're like, oh, I don't know.
I'm good blah blah.
But if you say, hey.
You know, I noticed that your social mediahasn't gotten any posts in the last 2 months,
let me create your next 10 posts for you andsend them over to you.
People are more likely to say yes to that andagreeing to help that.
(41:02):
And so I like what you said too about, like,helping people with something specific rather
than just, like, messaging people and say, whatcan I help with kind of thing?
Yeah.
Yeah.
Because it shows you're authentic and you care.
Right?
Like, it's really easy to say, hey.
What do you need help with?
Then the person who you're asking has to thinkAndrew they have to put mental effort into it.
It's, like, kind of unsure.
If you say, hey.
I'm really good at this.
Can I do this for you then?
(41:22):
Way easier for both of you.
Oh, I love that.
Let's talk about, or Syndrome Andrew.
So I know you mentioned that, you know, doingcode saves you 20 hours a week.
But do you ever present anything to any of yourclients, using AI and be like, oh, is it am I
(41:43):
really this good, or is it just AI making melook good?
Curious if you could talk about this a littlebit.
Yeah, dude.
It's it's funny because like AI is fascinatingto me.
The fact that you can take a bunch of data andpredict, for example, who's gonna churn or
who's gonna buy this product, or what's thenext best word to spit out in the sequence of
words just kinda blows my mind.
(42:04):
So there's almost always a sense of impostorsyndrome.
Like, oh, man, I hope this works this time.
Of course, it's gonna work this time becauseyou've tested it a hundred times.
Like, every time I demo my bots for Slack, andI say, well, you know, let's look let's ask a
question about this piece of information inthis this sheet or this document like thinking,
man, I hope this works this time.
So every time I do this, I'm like, there's acertain sense of fascination that it will work
(42:28):
reliably, and then there's a certain sense of Ireally hope I'm not coming across as a fake.
I really hope I'm being sort of genuine becauseI care about that.
I'm maintaining my integrity.
Because if you're trying to sell, yeah, I'm notthe type of person who can sell anything, so I
still ask them, oh, I'm like, this is what I'vegot.
I hope that this fits your needs.
Hopefully, it does.
So maybe other folks won't have problems withthat, but every time I sort of demo what I'm
(42:52):
doing and I talk about what I'm doing, it isone of those things where it's like, a little
bit too much hesitation, and that's somethingI'm trying to work at, but it's a long way with
me.
I don't wanna ramble.
It's like, yes, imposter syndrome is there.
Even though I've I've been working in datascience for 10 or 15 years.
There's folks out there who are way, way, waymore knowledgeable to me.
You, as a person, just have to stop comparingyourself to those folks.
You just have to say, this is what I could do.
(43:14):
Here's what I've done.
I you know, let's work together.
I promise it'll be a good experience.
And, I like what you said too, that you have toremind yourself, oh, yeah.
I have done this a hundred times.
I have I do know a little bit what I'm what I'mtalking about, and my track record can show it.
And like you said, too, you know, it's alwaysfunny and scary too, taking on new endeavors,
new projects, and and but, you just have tokeep going with what you know, but also trust
(43:36):
yourself to adapt accordingly.
So I like the way you phrased that.
And then I know we have a few minutes lefthere, so I have a few questions to wrap up.
What's the next end game that you're workingon?
It's funny.
I started last year was the was, a year that Ihad to shut down a startup that I had been
working 5 years on.
(43:56):
Andrew if if you had asked me at that point,hey.
Do you wanna do something else?
I'd be like, no way, man.
I am done.
And then, like, for most, like, mostentrepreneurs, I was like, after a few months,
I I said, well, I kinda have this idea.
I wanna see if this works.
I have this thing I wanna work on.
I realized I'm still I still have the bug tobuild something big.
So Locustiv is my attempt at that, but I'mdoing it very differently.
(44:17):
I'm doing it very, very systematically now.
So for me, the the big thing I'm trying to dois turn a lot piece of into a very, very large
company.
And I'm doing it in a way that is not hasn'tbeen done very often before, which sort of
taking it slow, not taking a lot of money untilI really have something that works, moving
very, very systematically.
So, that's what I'm trying to do now.
(44:39):
What what does a a large company mean or looklike?
What's the what's the vision for how how faryou see locus of going?
They're they're sort of the ideal.
And but then there's like, okay.
If that doesn't happen, what happens?
So, like, the ideal is go public and, you know,make it at least, multibillion dollar revenue
company, but that's something that maybe I willrun out of steam.
(45:01):
Maybe I won't have to do that.
Maybe 10 years go by, and I'm at a point whereI don't need that.
For me, a large company is something thatallows me to either draw quite a bit of money
from it on a year year to year basis or sellit.
I don't know what that number is, and Iactually don't wanna put a number on it.
You know?
Whether that's a 100,000,000 or a 1,000,000,000or even 10,000,000 or even 1,000,000.
(45:22):
I'll know it when I see it is the kind of thingthat I've that I'm saying.
But that's sort of what I'm trying to do now isbuild something sustainable and high value
Andrew just enjoy the ride doing it, but alsotry to take it as far as I can.
Wow.
I I love that, and that's, that's definitelyexciting, you know, bringing people in the show
with such big visions and aspirations and wannabuild a $1,000,000,000 company.
(45:45):
And Andrew going into the the mindset ofthings, when Twitter bought your company, could
you have quote unquote retired if you if youwanted to?
Yeah.
You know, I probably don't need to be workingnow, but I am also the type of person.
If I'm not working, I'm like, what am I doing?
Like, I've traveled a lot, which is great.
I've done a lot, but there's a certain sense ofjoy Andrew fulfillment I get from working and
(46:06):
building.
Specifically, I'm a builder.
It just happens that I build with tech andtools and software.
I get a lot of joy from building, and I get alot of joy from seeing people use the things I
build.
And so for me, it's one of those things where Ijust wanna keep pushing myself and seeing how
far I can go.
I love but but what, what helps you feel thatthat hunger to keep working and get that
fulfillment?
(46:27):
I mean, most people are like, oh, yeah.
I'll take the money and keep on chilling.
Like, what what what helps fuel you to staymotivated and keep going?
You know, I bet you I I I wish my wife knew.
I wish I knew.
I Weiss, like, you know, if she knew she wouldbe like, okay.
We can do this without you having to do it thisway.
So because I I'm still working.
I keep track of the time I spend now.
I it's not like 80 hours last week on thisstuff.
(46:47):
So I do some 13 or 14 hour days.
So I don't know.
I just know that that's who I am, and so I usedto fight it Andrew I used to deny it.
And then I was like, well, let's just embraceit.
So I kind of have a sense for why I am the typeperson who wants to keep pushing and keep
getting better, and I find a lot of joy frombeing able to level up.
It's like I played a lot of video games as akid.
(47:10):
But that's just innate to who I am.
And I kind of I've got some ideas of Weiss, butI tried not to dive too deep into that.
At this point.
That concludes another episode of rapidresults.
Remember to leave a review about something youlearn so others can share the keep being
unstoppable in your pursuit of the lifestylefreedom you desire, and we'll see you next
(47:34):
week.
Popular Podcasts
Bookmarked by Reese's Book Club
Welcome to Bookmarked by Reese’s Book Club — the podcast where great stories, bold women, and irresistible conversations collide! Hosted by award-winning journalist Danielle Robay, each week new episodes balance thoughtful literary insight with the fervor of buzzy book trends, pop culture and more. Bookmarked brings together celebrities, tastemakers, influencers and authors from Reese's Book Club and beyond to share stories that transcend the page. Pull up a chair. You’re not just listening — you’re part of the conversation.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com