July 9, 2025 • 48 mins
J.D. Trask, the Wellington-based entrepreneur behind global software success Raygun, is back with a new venture, one he believes could have an even greater impact than the internet itself.
In the latest episode of The Business of Tech podcast, Trask sat down with me to introduce Autohive, a platform designed to make AI automation accessible for every business, not just those with deep technical resources.
Trask’s well-established company, Raygun, is a quiet powerhouse in the tech world, providing error and performance monitoring for software used by everyone from Domino’s Pizza to HBO. With its behind-the-scenes tech and 93% of its revenue coming from exports, Raygun has flown under the radar in New Zealand, operating with a lean team from just off Wellington’s Courtenay Place.
But as generative AI exploded onto the scene in 2023, Trask saw a seismic shift underway. He described the electrifying moment he realised AI’s potential to transform business productivity.
"I cannot put down thinking about this, this is going to be a bigger revolution than the internet,” he remembers thinking. Autohive was born, and launched last week with a platform allowing anyone to make their own agents - with no coding experience required.
Tune in to listen to the interview in full - streaming on iHeartRadio and wherever you get your podcasts.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:07):
Welcome to the Business of Tech powered by Two Degrees,
the podcast where we deep dive into the story, strategies
and innovations shaping the world of tech from our perspective.
In our tairoa, I'm your host, Peter Griffin, and today
I'm excited to introduce a guest who's no stranger to
building global tech success from New Zealand, John Daniel Trask.
(00:31):
Many will know JD as the founder of Raygun. The
software is a service platform that quietly powers error and
performance monitoring for everyone from startups to Fortune fifty giants.
If you've ever ordered a Domino's pizza or streamed Game
of Thrones, chances are Raygun was quietly working behind the
(00:53):
scenes to keep things running smoothly. But JD isn't here
to talk about past achievements. He's here to unveil his
bold new venture, Autohive. It was launched with a pretty
effervescent party on the Wellington Waterfront a couple of weeks ago.
Autohive is on a mission to make the power of
AI automation accessible to every business, not just those with
(01:16):
a crack team of engineers. Imagine a platform where you
can build your own AI agents to tackle repetitive, time
consuming tasks that bog down your week, from invoicing to
customer care, all through a simple, approachable interface. Now, there's
been a lot of hype around so called AI agents,
(01:37):
so we're going to unpack what exactly an AI agent
is and whether Autohive have anything more compelling to offer
than some of the sort of multinational versions of agents
that are coming onto the market. In this episode, you're
going to hear about JD's journey from Raygun to Autohive,
why he believes AI is an even bigger revolution than
(01:59):
the Internet, The light bulb moment that led to Autohive.
That was JD realizing that if AI is to transform business,
it must be easy for anyone, whether you're a florist
and accountant or a retailer, to build and use your
own agents. Talk a bit about how Autohive's marketplace will
let you share or monetize your custom agents, and why
(02:22):
JD is determined to keep humans in the loop, ensuring
automation remains approachable but under control. Plus practical advice for
getting hands on with AI, building digital skills, and overcoming
the hype and fear mongering that too often clouds the
AI Conversation. Yeah, lots of great stuff in this discussion
(02:43):
and some of JD's sorts about what we need to
do to stimulate the New Zealand economy as well. Whether
you're an AI skeptic, a business owner looking to boost
your productivity, or just curious about where this AI tech
is heading, this episode is packed with insights you won't
want to miss it. Stick around to the end where
JD shares how you can get started with auto Hive
(03:06):
for free and why now is the perfect time to experiment,
learn and help shape the future of AI in New
Zealand and beyond. JD, Welcome to the Business of Tech.
How are you doing.
Speaker 2 (03:27):
I'm good, thank you, very very happy that the launchweek
is behind us. That was a lot of stress for
just getting it all sorted and get the product launched
and all of that. Yeah, that was a great event.
Speaker 1 (03:35):
Autohive launched on the Wellington waterfront, great party and a
lot of energy which the city sort of has been
lacking in the tech sector. But it came off the
back of some big, great conferences, the Electrified conference that week,
a big clean tech conference. In the middle of it,
you had this new startup launching auto Hive, and we're
going to talk about Autohive and depth, but I just
(03:56):
want to dial back a little bit to talk about
the company. You've devoted the last fifteen at least years
to Raygun, a company that's gone global, very successful, probably
a bit below the radar for most business people. Tell
us about Raygun, what you set out all those years
ago to achieve and what you have achieved.
Speaker 2 (04:17):
Raygun is not like a consumer brand, So for folks
who are thinking I've never heard of this before, absolutely
totally expected. You know. So what Raygun does is it
provides services through a SaaS platform to other technology companies
where they can track their faults and performance issues in
their software. And so, you know, we all sit there
(04:37):
and kind of cuss out our computers from time to
time when stuff just doesn't work. It's a fairly universal
problem and it's what Raygun helps with. So you can
think of Raygun as being like a black box flight recorder,
but the software code. So when something blows up, it
kind of gets the last message out to Sally, like, well,
this is what was going on at the time it
didn't work, and that is really helpful to the creators
(04:59):
of software, and so part to your question, you know,
why would kiwis not maybe know so much about us
as we're ninety three percent the export based, so most
of our money is we're pulling it in from overseas
and we're spending it here in Courtney Place, right, And
so we've got everything from small customers, new startups through
to Fortune fifty companies. So we've tracked for example, Domino's
(05:22):
Pizza Worldwide uses us, so if you're ever ordering a
Domino's pizza through the app whatever, and something goes wrong,
we collect the data, we helped them improve it. We
had Hbo originally as a customer and we tracked I
think it was eighty seven million concurrent viewers of the
finale A Game of Throne, So eighty seven million people
on Earth were sending data through our platform about how
(05:42):
long were they waiting for buffering? Was there any issues?
And so we operated a tremendous scale based just out
of here in Wellington. But that's our core business. That
was what we've been building for the last year ten years.
How did you get traction in the US?
Speaker 1 (05:57):
Initially that led to the Dominoes and HBO signing up
what was the breakthrough.
Speaker 2 (06:02):
Well, we've been selling products globally already under the brand
called Mindescape before that. That I started when I was
twenty three years old with a couple of other guys,
Jeremy Boyd and Andrew Peaters. They were the older, mature
people that I started the company with, and we sold
to developer tools there and then we built the raygun
product originally for Mindescape. It was our first subscription product.
(06:25):
I'd really admired what Rod had done with zero and
I thought, you know what, starting the month not at
zero is a good idea, and so we built a
subscription product around it and very quickly dominated all our
other products in terms of revenue. But simultaneously, because those
products are sold into software development teams, we knew from
(06:46):
literally day zero that we had to market this globally.
So you've got this thriving business. About twenty five people
involved in Reygun. Yeah, yeah, so have we. And this
is off a little bit at the back of the
AI story. We were actually pushing closer to sort of
sixty sixty five people a few years ago, and not
for you know, we haven't conducted rounds of layoffs or
(07:07):
anything like that, but when the AI stuff came along,
I kind of decided, well, if people were if people
were to leave, could we have a look at what
was necessary with technology to improve our output. I think
a lot of people kind of think that AI coming
along means they'll just do the AI version of their
current job. I'm increasingly not thinking that that is true.
(07:27):
I actually think it's an opportunity for people to really
think about what they like to do, and that suddenly
the how it's done is becoming a little less important
than it used to be. I'm not saying it's not
I'm not saying, you know that a person who, for example,
is never coded, is suddenly building operating systems, not at all.
But an example would be like these video creation tools
(07:49):
that have come out, and people are doing some amazing things.
And what I've noticed is it's not really the videographers
that are suddenly going, hey, I want to do this,
but it's it's seeing people that are genuinely quite caught
at their more creative that maybe it's like, maybe we
should have fifty times the number of directors now who
have creative talent that aren't blocked by needing an army
(08:09):
of people to deliver the output. Maybe there's more opportunities
for those types of roles to form. So anyway, backing
it up, so we let the head count generally sort
of drift a little down. We haven't seen any significant
slowdowns and productivity off the back of that. It wasn't
financially motivated, but it's becoming an attractor to see an
organization that kind of gets that this change is occurring,
(08:32):
and people's thinking, hey, well, at least go there. I'm
enabled to learn about how to move forward in this.
And I think that's one thing that is going to
catch a few organizations off guard if they kind of
continue to maybe play a wait and see approach, is
they're going to the people who don't really want to see. Yeah.
Speaker 1 (08:51):
Look, this is a trend we're seeing, particularly in Silicon Valley.
There's a lot of talk about this, about the fact
that you might have a company that can do one
hundred million dollars in revenue, but it does need more
than twenty people. Anymore, the expectation was as you scale,
you got more VC money or whatever you added, your headcount.
Speaker 2 (09:07):
Grew and grew and grew. It's not necessarily the case anymore. Yeah, well,
I think, and this is probably just me, but I
thought it was outrageous how much through Zerve all of
the product companies started to scale like they were services companies.
Like I totally understand if I'm building people out by
the hour, I need as many people as I can.
If I'm building a product that should scale out, I
should not need to keep scaling headcount as I scale
(09:29):
the product. The zero cost a duplication of software, that
is the benefit there. So I've always been a pretty
big proponent of smaller teams of highly capable people, very
empowered rather than giant factories. And in fact, you know,
probably a bit of a negative here, but I've been
saying this for about eighteen months now and I think
it's coming true, which is, if you want to avoid
(09:50):
the real risk of the downside of AI, do not
work in a company with tens of thousands of employees.
You know, you are by definition not wildly impactful, and
the ROI calculation is not going to be in your
favor when people are sitting there doing management by spreadsheets,
going oh, hang on a thousand people here, you know,
like that's some real money. I don't say that callously,
(10:12):
like I think it sucks, but I think that's what's
going to play out, and everybody should be sitting there
thinking about what's playing out.
Speaker 1 (10:18):
Well, that's playing out in the big tech companies over
the last two to three years. You know, we're seeing
four thousand people laid off in one go because the
spreadsheet being countered looking at that and going this is ridiculous,
as too many people in the books, and this is.
Speaker 2 (10:34):
Where you run the risk of zero sum game mentalities, right,
So there'll be a lot of people who freak out
kind of going, oh my god, the jobs are going away,
and it's like, well, no, the opportunity to create new
things are turning up, and you should be trying to
find what those are. So, you know, we were chatting
just beforehand about that Jeff Bezos quote where when you're
(10:54):
in a time of a lot of change, Jeff says
sort of the question isn't what's going to change, The
question is what's not going to change. And he gives
the example of Amazon where he's like, well, nobody's going
to turn around and tell me to ship things slower,
to increase the prices and all this. And that's where
I kind of look at this and I think, well,
at the end of the day, consumers are going to
continue to want to consume, you know, rightly or wrongly.
(11:16):
Human beings want more and more and more, you know.
And so you might sort of see an organization say
reducing head count but producing more. I don't think that
means that there isn't more organization starting up to also
produce things, you know. And so it really comes down
to that mindset. And I do worry a little bit
that the Kiwi mindset is too zero s. If somebody
(11:39):
else wins, there must be a loser. Therefore, you know,
we should penalize the winner. And it's like, no, you know,
we produce enough food to feed forty million people from
New Zealand. Maybe we could look at the next category
down and say, how did that actually served you know,
five x the number of people that will bring in
more money. That will be all that lifts the country
up is to bring more money into the country. And
(12:00):
you know what really bugs me about this, I've been
talking about this for years and now it sounds like
I'm talking Donald Trump's book, you know. I like, hey,
we shouldn't have a trade deficit. It's like when New
Zealand really shouldn't have a trade deficit, and it's had
one for nine out of the last ten years. Right,
we are progressively getting poorer and therefore we're asking a
lot of questions now, which are like, well, who can
(12:21):
we take the money off then if we can't figure
out how to earn it, And it's like no, no, no, no, guess,
let's work out how we can become hyper productive and
sell a whole lot more stuff to the world. That's
going to lift all the boats. You're going to increase
your tax take, you're going to increase people's quality of living.
All of these things come from business. They all come
from the private sector, and ideally in an exporting situation.
Speaker 1 (12:42):
Yeah, and look, we've got a government going for growth,
so that the rhetoric of the political rhetoric is in
that camp. You know, there's more momentum now, but our
execution has not been great in our trek record of this.
We have great companies who scale and employ people. You know,
take Sect. One hundred thousand dollars plus salaries. It's just
(13:03):
how do we do more of it? I guess is
the question.
Speaker 2 (13:05):
Yeah, and look there's lots of ways they could you know,
there's plenty of things we could do around improving our
R and D tax settings, for example. And the problem
is is that the tighter things become, you know, the
less options there are available. And I don't think many
Kiwis realize how tight they've already become. And that was why,
you know, part of the rhetoric at our auto Hive
(13:29):
launch is just sort of say, look, we can all
set around complaining and look, I've complained more than most
you know about about Winger's the government or the local
government going to step in and do something. But the
reality is there's actually society that does all of the work,
you know, and that includes the people in government. I'm
not sort of us them in that regard, but it's like,
(13:52):
why do I worry that too many people have since
at least since Az a kid has sort of shifted
their mentality here that they leave home and the government
becomes their parent, and it's like, what the hell are
you doing about? This is your shot to go and
make your impact in the world, you know, run out
there and do something amazing. Stop whinging about. I mean,
(14:13):
I have that issue when I see people saying there's
no jobs and I'm like, well, I've had exactly zero
people come and knock on my door saying, hey, can
I know your lawn's going to paint your shared Like,
I'm like, you know, where is the agency here? You know,
it's just no, no, the government should pay me more
money while I'm not working. Yeah, I'm not so sure
of that. I think we want and I think that
pioneer spirit is actually was the Kiwi spirit. I'm not
(14:36):
saying that to hammer people. I'm just sort okay, let's
bring that back. That was a much more fun attitude
to sort of look at the world through. Yeah.
Speaker 1 (14:44):
Yeah, So the other night you related to quite an
interesting story about early twenty twenty three. I think it
was the rise of generative AI. Chat ChiPT had come out.
The previous novembers would have wowed everyone, and you're sort
of planning the future of Reaguan and sitting down and
this time around was a bit different. You're like, I
(15:04):
don't really know where this is going. We really need
to think about what the impact. You said, it was
an electrifying moment playing with all this new generative AI
stuff and then thinking about how do we apply this
to Reagan and then thinking is there a business here
helping small and medium sized businesses use AI. Tell us
about that sort of that moment, that light bulb moment,
and what it meant for you as as an entrepreneur
(15:26):
and a business.
Speaker 2 (15:28):
To set the scene a little, I was in my
early teenage years when the Internet sort of came along,
and I was a huge nerd, so I loved playing
with that, you know, and I was cognizant at the
time that I was probably too young to sort of
make a go of building a business on the internet,
if you will, at that point. Now, admittedly I did
(15:49):
start building a business on the Internet when I was
twenty three, but in my teenage years, I didn't really
feel like I was in the position to really exploit
what the Internet was enabling. I spent a lot of
time reading about Microsoft and all the tech giants at
the time. You know, There's a great book called The
Microsoft Way that I enjoyed, and they talked a lot
(16:10):
about how those organizations were trying to grapple with this
shift in technology and so when this when the Internet,
so when the AI stuff sort of kicked off, this
to me, I sort of immediately recognized all the same
feelings and I think, you know, this probably applies to everybody,
but the older you get, you know, the more you
(16:30):
realize that your first inclinational gut feeling was absolutely correct,
you know. And that was why I was like, I
cannot put down thinking about this. This this is going
to be a bigger revolution than the Internet. It's going
to be bigger than computing. It's going to be the
biggest wave that mankind has ever seen. And if I
go back, I talked to the team. As you said,
(16:50):
I kind of went in and I didn't have a
clear vision for the year. I just knew that this
thing was colossal. I hadn't had enough time to play
with it. I hadn't formed strong enough opinions, And so
we spent the time talking about the nineties and what
did we see from the Bellgates Internet tied awave memo.
A few people realize this, but like Jeff Bezos was
(17:11):
working at a hedge fund before he started Amazon, and
the story goes that he's researching different industries and he
sees that the Internet is growing at twenty four hundred
percent per annum, and he just looks at this as
like nothing grows this fast. You know, something is going
on here, and we're seeing that, but potentially in order
of magnitude even bigger, you know, and adoption rates right now.
(17:32):
So I had this option where I was like, hey,
I've got this successful business. It's doing great stuff. That's
really cool. I see this giant thing happening. I definitely
need to apply some of that to what I'm doing today.
But also I didn't want to be completely handcuffed to
an old way of doing things. And so hence auto
Hive both as a separate brand with the intention later
(17:54):
of spinning it out as its own company. So it's
not pivoting Reagun Raguan just happens to be birthing this company.
And as you say, so we built. We sort of
had that time in early twenty twenty three. I had
the meeting with the whole team and was sort of saying, hey,
this is what's going on. We decided we then needed
to learn a lot more. So we had our AI
(18:16):
week in May twenty twenty three where we stopped all
work except for doing customer support. You know, I don't
want to leave a customer in the lurch, And in
the first four hours we got everybody to do our
engineering onboarding exam. So when we're hiring an engineer, you know,
we give them a test. And at the risk of
sounding a bit arrogant here, we have really really good
(18:37):
engineers at Reygun. Like I say, we scouted at eighty
seven million concurrent users for HBO. You know, I'm like,
look at trade meet, don't have to have everyone in
the country visit to just hit five point five million.
That was only one of our customers. So we know
how to scale, we know how to build international, and
so I wanted to so we built a range of
(18:59):
age to help us in our own business. These are
things like improving our AdWords, analyzing and bound trials, anything
we could kind of look at and think, what is
some toil that happens in our company that we could
take away, And so we built these agents. And the
key learning I sort of had was like, these are
amazing they have had they have improved our business, they
(19:21):
have allowed us to make better decisions. This is great,
but they're also really hard to build, and that you
needed this crack team of engineers to put this together.
And so that was exactly what you said earlier. Was
sort of the epiphany was like, hang on a minute,
if this technology is supposed to be running in every
single business out there and you need to crack squad
of software engineers to get any value out of it today,
(19:44):
that's not going to work. And so we set about saying,
how could we build a platform or product for the
everyman basically or every woman to say, you know, how
does the local accountant or the florist or the retailer
like take advantage technology. Now we're all kind of going
into chat GBT and asking questions and getting stuff, but
(20:06):
at the risk of over laboring the analogies. I think
of that back to the Internet is like when you
got email and you're like, you know what. Email was great.
You didn't have to go and spend twenty five cents
to send a message, and it didn't take three to
five days to get there, you know. So we suddenly
had free mail and that was kind of the hook
for the Internet, but it wasn't actually the major value
of the Internet. And I think with AI, what we've
(20:28):
seen is that the chat GPT is like the hook,
but the real value is coming in these agentic systems.
So appreciate the listeners who have probably gone this far
and maybe wondering what the hell and AI agent even
maybe we should explain it. Yeah, but yeah, let's talk
to that so and you know one of the things,
and like, if you're listening to this and it still
doesn't make sense, please send me an email at Janie
(20:50):
Trusk at autohive dot com and I'll try it again,
because I do value being able to take technical concepts
and try to make them approachable. But when you were
working with say chat GPT, and you ask it a question,
you might say, you know, what was the best test
score in the rugby last year, and it gives you
an answer. That's cool. You might ask it something that's
happened today and you'll see it go and do a
(21:11):
Google search for example. That is what's called tool use.
So imagine the AI has been given a hammer, but
the hammer is called Google search, and so it kind
of goes, well, I've got this in my tool belt,
should I need it, And you've asked me for something
that's you know, clearly happened only today, so I'm probably
going to need to go get some more information, so
it'll use that. Tool agents are really, in my opinion,
(21:36):
are just the tool using AIS. And so the question
then becomes how can I give it some extra tools
that would make sense to do more? So in our case,
for example, we use HubSpot for our help desk and marketing. Well,
let's give the LEM a tool that lets ask questions
of HubSpot and our help desk, and maybe it could
even send messages through it, you know, those sorts of things.
(21:58):
And so then we can have an agent that says, well,
I'm watching the help desk and I saw that our
new ticket came through and I read it, and you
know what, I actually have the tool to put the
draft reply on there, so that when you know, JD
or somebody on the team has turned up to work,
they can just review that it makes sense and clicked scent.
And so it's allowing these the AI to sort of
(22:19):
quote unquote break out of the box. How does it
start to interact with the world around us, even if
for now it's software based interaction. And so that's where
we have these examples that we you know, we had
a demonstration I think on the night playing in a
video where I made a simple agent didn't take more
than a handful of minutes, where I said, you're going
to help me find properties and I said, we're in
(22:42):
New Zealand, you know, consider local councils, all this sort
of stuff. You just put that text in the prompt right,
which is the main instruction. You write that in English.
And I gave it the ability to search the web
and read web pages. That was all I did. And
now when I say, hey, I'm looking for a property
I wanted to spend you know, so one point two
million dollars in Mount Verk it needs to be in
(23:04):
you know, the school range of this thing. And you
watch this agent go off and spend the next sort
of ten to fifteen minutes browsing real estate, dot code
and Z trade me cross. It builds up a list
of like properties that's going to suggest. Then it goes
and checks the council websites and it gives you the
questions you might want to ask the real estate agent.
And this is just one simple example you know that
(23:26):
you can put together and then you start looking at
all these for business. So like I say the question,
I always always ask people, and I encourage the listeners
to think about this is just what is a task
that takes you one hour or more every week or month?
You just fre again hate everybody has a few and
(23:47):
voicing exactly. You have these things where you kind of go, gosh,
if somebody took that away, you know, I'd go and
focus in these areas. And that's that's kind of a
little bit what I think about when it comes back
to that disruption thing. It's like you're really ina blling
people to focus in the areas they actually want to
put the effort into, you know, and play to those
strengths and take away the toil work that's not actually
(24:08):
much fun for anybody.
Speaker 1 (24:10):
It's those time consuming tasks like invoicing or customer care
and automating that. But you've built a platform sitting behind
at is the large language models. We're using it in
a consumer sense, the open ayes models, presumably Claude and others.
So you're using the best large language model to apply
(24:31):
to a particular task, and you've built a beautiful interface
that makes it really simple to build your own agents.
Speaker 2 (24:38):
Yeah, one hundred percent. So first off, it's I'm sure
the folks that are paying a bit of attention today,
I see this. You know this week Claude will be
the best. Next week CHATBT the week after that, you know,
Elon might finally ship something you know, all of these
things keep playing out, and so we want to sort
of say, hey, it's kind of it's somewhat irrelevant. You
(24:59):
should be to use whatever the best thing is the
time through our platform. You're right, The entire UI and
everything we're working on is to try and make something
that is not for nerds. We want this to be
for people who say I want to jump in and
as you say, maybe I want to automate a bit
of invoicing. I want to be able to like toggle
on the zero integration, you know, and say, hey, give
(25:21):
me a more readable version of my pen now, or
just describe what the biggest risks are or things like that.
You want to make people so they're empowered to do stuff.
One of the things we're trying really hard to do
as well is avoid all of the acronyms. Like you know,
I've been in the tech industry for a long enough
time that that's usually the first place we lose people. API. Yeah,
(25:42):
my API talking to the RAG system so that you know,
my AI can You're like, what the hell are you
talking about? You know, And so often the tech industry,
I think rightly gets a bit of an eye roll
because it almost feels like we set out to make
things more complicated than they should be. So yeah, we're
trying to make something where it's hey, leave the really
nerdy stuff on the back end to us. We'll figure
(26:03):
that out. I probably wouldn't talk about this on an
international podcast like mono Hive will be an international product,
but I want to have that impact on our local
businesses in New Zealand. We know we have a productivity issue.
We know this technology is coming, there is now something
home growing for it, and so we get a huge
amount of value from people just feeding back one are
the bits that suck. You know, it might sound really
(26:25):
weird to say, but it's like when you launch version one,
it's the worst state it's ever going to be in
for the consumer, right, so we want to learn as
fast as we can. You mentioned HubSpot.
Speaker 1 (26:34):
You've got relationships with a number of box Gmail for instance,
So this is very much you know, it relies on
you getting that often sensitive but high quality company data
that's sitting in various data repositories. We've heard so much
from the big tech companies that have gone all in
on agents salesforce. In particular, Mike Benioff is betting the
(26:57):
company basically on agents, Microsoft, SA, all of them, and
their whole thing is bring your data to us, keep
it really clean on our platform, and it will allow
you to do all of this sort of stuff. How
do you approach that because you haven't got hold of
everyone's data in one place. Are we back to the
so called APIs to get that data into autohive?
Speaker 2 (27:19):
Yeah, so let's break that question down. So firstly, all
the big guys want to be the hub in the wheel,
so they want all the data to bang on, you know.
And I always think if Mark Benioff really believed in
what he was doing, he wouldn't be selling three million
dollars with a stock every day, which he does. So
you know, we've been a Salesforce customer, and to be honest,
(27:40):
I think we're going to see some bad behavior off
the back of AI. For example, Salesforce have recently said
to some companies that they will prevent them from integrating
into Slack because they don't want AI companies to be
able to get that data. I think I own the
damn data, but what my team are talking to talking
about in Slack, and so I think we're going to
actually start to see some of these particularly big companies
(28:02):
start becoming quite bad actors about who actually owns company data.
And I don't think Salesforce is actually doing a very
good job of demonstrating good behavior. Now that's a very
Salesforce specific answer for a moment, but ultimately I think
that as much as anybody would love to sort of
own a business where every party puts one hundred percent
(28:23):
of their work on your system, it is it is
naive and foolish. I think the reality is we actually
work in a mixed world of various systems. You know,
I'm not going to suddenly move my accounting of zero
to something that Salesforce makes just to improve an AI.
I want the AI to be able to talk to
zero and do a great job. And then what we're
seeing out of a lot of the foundation or the
(28:45):
big call them the big guys if you want. And
this is a big differentiation from autohyper auto Hive should
be the glue between all of the systems that you've got.
It should be the interface that you go and use
to work with. Primarily, it should be nice and joyful
to use. Matter that you're pulling data out of Google Drive,
out of zero out of Gerro, out of HubSpot, and
we bring all of that together, we take care of
(29:07):
the messiness of how to do that behind the scenes
for people. And I think when you look at some
of these eye watering valuations on these AI companies, look,
maybe I'm just under and I already think I'm estimating
a pretty big size and scale of AI, but maybe
I'm underestimating. But some of those valuations seem to be
(29:28):
predicated on the idea that those AI companies completely own AI.
And I'd be remiss if I didn't sort of highlight this.
You know, we have some folks even in our business,
they definitely believe as an AI, but some of the
first call bs on the hype and I'm absolutely one
of those people too. We've talked a bit about the
(29:49):
potential around job losses and things like that, but the
reality is today it's about automating a handful of hours
every week in toil right. Technology very well may improve
over the coming years to make that more impactful and
more cavble. I have no doubt of that, But today
it's about save yourself a little bit of pain overall
(30:10):
in there. I don't think people want to be starting
that journey with I have to pile it all on
to one megatech.
Speaker 1 (30:17):
So it's been out there, you know, only launched it
maybe a couple of weeks ago. But what's been the feedback,
like in some of the use cases that are starting
to bubble up with auto high.
Speaker 2 (30:26):
Yeah, so we actually launched it silently last Tuesday, and
then Wednesday was there, let's start making some noise on it.
So it's not actually been out a week yet, so
we're pushing four hundred odd folks who have now come
in and tried it out, which is really cool. We
can see folks have made a range of agents. I
(30:47):
will say we do full This sort of touches on
an earlier question, but there's full encryption across all of
the conversations. We can't actually read, you know, I can't
sort of sit there and go, oh, Peter, I asked
this question. That's all lockdown, and we're very mindful about
the trust and security side of things. But we can
see people making agents that they are sort of setting
(31:08):
up these different tools to run things. And I have
had a few people citing the examples like the property
one that I made, which is how I know exactly
what that was doing in there. The main piece of
addressable feedback that we'll be tackling over the next like
I said a couple of months, is that marketplace piece.
So it's great to come in and you can use
(31:30):
all of these l elems. You can use Chantypete, Claude,
Gemini all in here, and we actually have a you know,
it's free to start. You know, we have ten dollars
worth of credits on there, so you can just use
it for free to begin with. Then we have a
paid tire where we go to eighty dollars a month,
but this is not per user. So if you've got
a company of one hundred people, pay eighty bucks and
(31:51):
put them in there. Now, there is some usage pricing
that can go on top of that later on. So
if you do have a few people that are really
hardcore and they're using a lot, just be the price
of the tokens plus a small margin on there. Now,
why this is useful is because, and I can say
this even across the tech industry for those that are
listening that maybe aren't in tech companies, the adoption of
(32:14):
AI is very sort of in disparate groups right now.
So even in software companies, you're sort of seeing twenty
maybe twenty percent of people are really leaning in and
discovering what's possible. There's about fifty percent of people are
sort of, you know, maybe asking a handful of questions
each week, you know, and then twenty five percent of
(32:34):
people who are just not at all engaged. The problem
with that model is, obviously, if you're the CEO, you
want to get your team up to speed. You want
to roll out all the tools. If I was to
go and get Gemini, Claude and chatchby t for a
single team member, that would add up to sixty US
dollars for that one team member right now, if that
team member is in the group that isn't really using
(32:56):
it a lot, I'm going to start getting frustrated that
I'm paying for something that's not being you. And so
that's why we've tried to make the model one where
it's like, yes, there's the starting price, but that gives
you enough credits to sort of not hit the wall
too quickly. But then you're only going to scale it
based on if you're getting the utilization. And that's where
I think ultimately we want to be aligned with the
(33:16):
customers is to say it's in our interests and your
interests that we're delivering creating agents that are delivering real
business value. Because if I can get ten thousand dollars
worth of ROI out of two hundred dollars worth of spend,
I will do that deal every day of the week.
Speaker 1 (33:32):
Yeah, and for a small so you've got a ten
person business or you know, twenty or thirty person business,
eighty bucks a months a month. If that's spread across
various agents, because you can set up multiple agents and
I see you know, they even have their own avatars.
So increasingly we're going to be working along these different agents.
(33:53):
You'll have the agent for invoicing, you'll have the one
for customer service, and in auto hive you'll see the
progress where what they're doing, what they're reporting, where they're at.
Speaker 2 (34:02):
Yes, one hundred percent. So for those of you that
are listening, imagine a user interface that looks a little
bit like Slack or teams, but we say half of
the users are actually smart agents. So you might have
like a governance one that understands your shareholder agreement. You
might have I've made one that advises me on my
personal stock portfolio. You know, it can read that data
(34:24):
and it just sort of says, hey, you know, you
might want to rebalance this way, or here's the earnings
reports coming up, those sorts of things, and so what's
kind of fun about it is, don't think about bringing
autohib into your company, like, okay, cool, we need to
mentally spend you know, one hundred thousand dollars this big
bang deployment. We need all these agencies. He no, think
(34:44):
of it as like you could go and sign up
as the CE even you know, you should be able
to make a couple of agents to do something simple
and then you can invite your team in for free.
They don't cost anything more. It feels a bit more
like an infinite box of Lego, you know, where it's like, cool,
build something, just start to explore, you know, and then
you'll start realizing, oh, I could build even fancier things
(35:06):
and fancier things. And we found this even in building it. Right,
So we would have a discussion internally where somebody said, oh,
how could we build an agent that does X, Y Z,
And I was like, well, I'd probably do it. ABC
one of the younger team members was like, actually, we
could just have the agent talk to another agent and
do this, you know, And that's the power of the
(35:27):
creative thinking. Like a lot of us that are a
little bit older kind of have already got an idea
of how to work, and these things are quite a shift.
And so that's why even with an auto I've mentioned
the easy to use you I trying to make it finally,
sacronyms and all of that.
Speaker 1 (35:42):
I guess the big promise from the Benioffs and that
is we've moved from this information generating ability with generative AI,
which is incredibly useful, through to doing stuff on your behalf.
And I remember twenty eighteen sitting at Google iOS conference
where they debut Google Duplex, which gone nowhere. But they
made a call or this AI agent made a call
(36:04):
to a salon and had a conversation. This was an
AI bot having a conversation with the salon owner to
book an appointment. So the point where we get to
aim making decisions on our behalf and adjusting if the
salon owner goes, we're booked out this week, you know
what about next week? Going and consulting your calendar to
see if you have a slot next week. How far
(36:26):
away are we from that sort of stuff, we're pretty
much there there.
Speaker 2 (36:30):
Yeah. I mean the calendar example, just on the auto
hive side, I played with this. I just attached my
Google calendar and I started asking it questions like what,
you know, how would I optimize my week and things
like that. So the pieces are there, They're all on
the table. I often. That's why there is just fundamentally like,
the amount of opportunity in AI right now is just insane.
(36:53):
Any direction you look into is just a blue ocean
of opportunity right now. That I used to think when
I was building reagun, Gosh, wouldn't it be nice to
be in the blue ocean rather than like competitive field?
And now I kind of take the viewing like, boy,
it's quite disabling to have infinite opportunities in all directions
(37:14):
and how you kind of keep that focus, Which is
why we keep coming back to that SMBs, which by
New Zealand standards pretty much everybody you know, and ease
of use, you know, that's what we're going for. But
the ability to glue these things together to create these
complete experiences is there absolutely today I have to use
(37:34):
the iPod. It's a good example of that, which is
The original story of that was that one of the
hard drive makers I forget which one, had actually told
Apple that they were building this very very tiny hard
drive and they had no idea what to use it for.
(37:55):
And it was Steve Jobs and Johnny I've that kind
of sat down and we're like, well, hang on, if
we can put this much memory into its device of
this size, you know, we could do this, this and
two hundred songs. Yeah. But this is the thing I
always think is most interesting about a lot of innovation
is how frequently you kind of look back and you
realize all the pieces were on the table. It just
took a person working out the right way to put
(38:17):
them together that they make these phenomenal products. And I
feel right now like there is a tremendous amount of
pieces on the table, and we're going to spend years
building the ability to leverage it. Yeah.
Speaker 1 (38:29):
So what's the plan for the you know, the remainder
of twenty twenty five. You're really building this with the
local focus first, but clearly trying to prove some use cases,
get some feedback from customers, very with a view to
very quickly taking it international.
Speaker 2 (38:46):
Yeah. Absolutely, So there's nothing that prevents international people signing
up today. And because we do have the background in
Reagun like Reagun does actually have people outside in New
Zealand too, so we've already got you know, like you know,
we're already set up for account in the US and
we've already got the tax stuff and security and all
that sort of stuff is already in place, which is
good between now and the end of the year. The
(39:10):
main one is that marketplace capability. And I don't want
people to think of the marketplace is purely like I
have to go spend money. As you say, there could
be a lot of free agents on there, but it's
that ability for people to almost pick from a menu
rather than have to make something. You know, I really
want to dial that up to eleven. So those use cases.
We do have more than a thousand use cases listed
on the site, but just generally, how do we lower
(39:31):
that barrier to entry. So between now and Christmas you'll
get to a point where you might search for say
you might kind of go agent to maybe do zero invoicing.
You know, you want to have that come up and say,
well here's the page on the autohib site with you know,
there might be three of these agents. They should have reviews.
Click on that. Oh yeah, I like to look at
that one, but click get sign them with Google and
(39:52):
be dropped straight into talking to that agent you know,
able to assist you get people into the ecosystem. There.
One observation is just that we well have to put
a lot of work and this is a good thing actually,
and to really leveling up people's understanding of how to
even prompt these tools, there is definitely a real skill
(40:15):
in that. I think again, just as a tip maybe
for the people listening is I've often found that a
mistake people make is that they give too little context,
so they only maybe give one or two sentences, which
works really well if you say, to say, chat GBT,
you know who won the presidency in nineteen eighty four,
that's fine, But if you want to actually do something
(40:36):
somewhat complex, you've got to think of the prompt as
sort of setting the parameters in which the AI should
adhere to and then be open to accepting all possible outcomes. So,
you know, this is where we get the crazy stories
from the movies where it's like, you know, is it
two thousand and one Space Odyssey, where it kills everybody
(40:58):
on the ship. Now, you just said you needed the
people over here, and so they had to be alive, right,
And it's like that's the bit that trips people up
and it sounds like we're being insulting to AI. But actually,
if you think about it as just a general intelligence,
if you didn't say these things, you know, like, how
(41:19):
would it know? And so oftentimes when I see people struggling,
it's either putting in two little context or that're trying
to micromanager, like they literally want it to spit out,
like they know to the letter what they want in
the outcome to be. And it's like, don't do that either,
like you wouldn't you wouldn't expect that when you're managing,
say a junior team member either it's not going to
(41:39):
be a good way to work with them.
Speaker 1 (41:41):
Yeah, And there's just been so many reports. Microsoft just
published one about the productivity potential of AI, you know,
so I've covered so many of these reports, but it
just seems to be a gap between what's possible in
the short term in the next five to ten years,
and are ability to exploit that. And it really comes
(42:02):
down to digital skills and people having the confidence to
actually get hands on with these tools.
Speaker 2 (42:07):
We've got a lot of work to do there. Yeah,
at a fundamental level, totally agree with that. I do
think to your point, there's a lot of reports and
a lot of noise coming out of these big companies.
I think they are over hyping things a bit. Simultaneously,
I think they are also fair mungering way too much.
They're creating them. I mean I saw a thing the
(42:27):
other day where one of them was like eighty five
percent of the time the anthropic model was blackmailing people.
I'm like, you would have had to go out of
your way to prompt that thing to do this, and
you're trying to scare the bejeebus out of people. You
know this, This is a bit crazy. So, as a
few people sort of highlight, a lot of those big
(42:47):
companies are really pushing to try and create a regulatory
capture environment where only a handful of them are allowed
to operate. Yes, and it's it's not cool.
Speaker 1 (42:56):
Are you bootstrapping this out of your revenues from Reagan?
Speaker 2 (43:00):
Yeah? So reagun is a fairly well established business. So
at the moment it's being funded through reagun. And once
we've driven up revenue. Like I say, the intention will
actually be to spin it out as a standalone company
rather than keep it as a business unit. Reagun itself,
like I said, is doing fine. It's still very valuable tool.
(43:21):
We're going to continue to add AI capability so that
we already have some in there overall. So yeah, that's
the that's the approach we're taking at the moment.
Speaker 1 (43:31):
Yeah, so it's really about getting those use cases for
auto hive. People come back going Wow, this is saving
me a lot of time and energy, and then applying
that hopefully to others.
Speaker 2 (43:42):
Yeah, this is soon one hundred percent. And your point
earlier about you know, the over hyping from the big guys.
That's why I say start rather than sort of having
this existential crisis of like if I bring this in,
am I implying that I don't need stuff, and like, no,
you're not. You're probably going to automate some of the
more boring things that happen in your business, and you're
(44:04):
probably going to find the wow factor is really just
to be able to say that some of that work
has been taken off your hands and there The other
thing I'd like to just address is I think we're
drawing to a close here, as you mentioned as well,
having these things do stuff on your behalf, and I
want to address that just because I'm sure there's some
people they're thinking, gosh, there is no way in heck
(44:28):
that I want this thing to do stuff on my behalf,
you know. So what we've done is we have what
is called human in the loop. And this is somewhat
common or becoming more common, but it's the idea that
you can choose the actions that need your explicit approval.
So the email example I have was I attach it
to my inbox and I could say what's my writing
(44:49):
style or give me some tips on improving of you know,
who's my most annoying person? Email? But I can choose
that to send an email will require that auto Hive
actually prompts me and says I'm going to send this yes, no?
Or do I want to interject and say maybe tweak
the message this way first and show it to me
so you have absolute control on what you're going to
(45:12):
allow it to do autonomously. That's how a lot of
the very cutting edge tools are operating. And my vision
for this and you asked about between now and the
end of the year, and I talked about the use
cases with the marketplace. We'll also have a mobile app
by the end of the year. My dream for this
is I want to wake up in the morning and
I want to have had these autohive agents have drafted
(45:33):
or my email replies. I want them to have read
my calendar and see that I have a board meeting
and I've got this meeting at this meeting, and I
want it to prepare all the work as far as
those agents can take it, but they won't necessarily have
completed it, so you know, it might have drafted, say
five email replies. I want to be able to get
out of bed. Maybe I'm still lying in bed, you know,
look at my phone. Yes, yes, no, no, yes, you know,
(45:54):
I'm like, you know what, before I've even got out
of bed, I've actually just kind of cleared a bunch
of this. Yeah, all's work. That's the goal, you know. Ultimately,
it's about trying to free people up to focus on
the stuff they actually enjoy doing. I don't. I don't
just like the people who send me emails, but running
emails is not my top most interesting task of the day, or.
Speaker 1 (46:16):
Spending time on in New Zealand's website, flicking through looking
for flights and all that sort of thing, if I
can get the agent to just come back to me
and go, this is what I found. Do you want
me to go ahead and book that for you? I
think a lot of people be comfortable with that.
Speaker 2 (46:28):
Yes, yeah, I was using that a fair bit when
I was traveling in the US, you know, like I
want to be in this area, I want to spend
this sort of money. Find me some options, you know,
just go and have a look.
Speaker 1 (46:37):
Yeah, great, Well, thanks JD for the rundown and all
of that and on agents and where they're going. Good
luck with auto hive and as you've said, you're happy
to help anyone who wants to see it one up.
So that's a great opportunity for anyone listening.
Speaker 2 (46:50):
Yeah, one hundred percent autohive dot com. It's free to
sign up. There's some free credits so on there as well,
and you can talk within our community and we're happy
to help. Ultimately, I just want to see you said,
I'm elevating on this and doing better with AI because
I think it can actually help us with some of
the challenges we face today. That's it.
Speaker 1 (47:13):
For this episode if the Business of Tech. A huge
thanks to JD Trask for sharing his vision and some
practical wisdom on building with AI, not just for tech giants,
but for every business ready to take the leap. And
if you are ready to do that and see what
AI agents can do for you, check out autohive. Sign
Up is free at autohive dot com. The credits are
(47:36):
on the house the first ten dollars worth anyway, and
JD and his team are eager to help you get started.
At businesses dot co dot enz. In the podcast section
you'll find my show notes with so many other things
that JD mentioned, for instance, the Bill Gates famous Internet
title Wave Memo, which is worth a read. Find them
in the podcast section at businessdes dot co dot anz.
(47:59):
Don't let the future automate you out, that's the real message.
Get ahead and automate the future yourself. Next week, a
rising star of our space sector attempting to solve a
really big problem facing the companies that are launching ever
more satellites into orbit around the world. That's next Thursday.
(48:19):
I'm Peter Griffin. I'll catch you then,
Welcome to the Business of Tech powered by Two Degrees,
the podcast where we deep dive into the story, strategies
and innovations shaping the world of tech from our perspective.
In our tairoa, I'm your host, Peter Griffin, and today
I'm excited to introduce a guest who's no stranger to
building global tech success from New Zealand, John Daniel Trask.
(00:31):
Many will know JD as the founder of Raygun. The
software is a service platform that quietly powers error and
performance monitoring for everyone from startups to Fortune fifty giants.
If you've ever ordered a Domino's pizza or streamed Game
of Thrones, chances are Raygun was quietly working behind the
(00:53):
scenes to keep things running smoothly. But JD isn't here
to talk about past achievements. He's here to unveil his
bold new venture, Autohive. It was launched with a pretty
effervescent party on the Wellington Waterfront a couple of weeks ago.
Autohive is on a mission to make the power of
AI automation accessible to every business, not just those with
(01:16):
a crack team of engineers. Imagine a platform where you
can build your own AI agents to tackle repetitive, time
consuming tasks that bog down your week, from invoicing to
customer care, all through a simple, approachable interface. Now, there's
been a lot of hype around so called AI agents,
(01:37):
so we're going to unpack what exactly an AI agent
is and whether Autohive have anything more compelling to offer
than some of the sort of multinational versions of agents
that are coming onto the market. In this episode, you're
going to hear about JD's journey from Raygun to Autohive,
why he believes AI is an even bigger revolution than
(01:59):
the Internet, The light bulb moment that led to Autohive.
That was JD realizing that if AI is to transform business,
it must be easy for anyone, whether you're a florist
and accountant or a retailer, to build and use your
own agents. Talk a bit about how Autohive's marketplace will
let you share or monetize your custom agents, and why
(02:22):
JD is determined to keep humans in the loop, ensuring
automation remains approachable but under control. Plus practical advice for
getting hands on with AI, building digital skills, and overcoming
the hype and fear mongering that too often clouds the
AI Conversation. Yeah, lots of great stuff in this discussion
(02:43):
and some of JD's sorts about what we need to
do to stimulate the New Zealand economy as well. Whether
you're an AI skeptic, a business owner looking to boost
your productivity, or just curious about where this AI tech
is heading, this episode is packed with insights you won't
want to miss it. Stick around to the end where
JD shares how you can get started with auto Hive
(03:06):
for free and why now is the perfect time to experiment,
learn and help shape the future of AI in New
Zealand and beyond. JD, Welcome to the Business of Tech.
How are you doing.
Speaker 2 (03:27):
I'm good, thank you, very very happy that the launchweek
is behind us. That was a lot of stress for
just getting it all sorted and get the product launched
and all of that. Yeah, that was a great event.
Speaker 1 (03:35):
Autohive launched on the Wellington waterfront, great party and a
lot of energy which the city sort of has been
lacking in the tech sector. But it came off the
back of some big, great conferences, the Electrified conference that week,
a big clean tech conference. In the middle of it,
you had this new startup launching auto Hive, and we're
going to talk about Autohive and depth, but I just
(03:56):
want to dial back a little bit to talk about
the company. You've devoted the last fifteen at least years
to Raygun, a company that's gone global, very successful, probably
a bit below the radar for most business people. Tell
us about Raygun, what you set out all those years
ago to achieve and what you have achieved.
Speaker 2 (04:17):
Raygun is not like a consumer brand, So for folks
who are thinking I've never heard of this before, absolutely
totally expected. You know. So what Raygun does is it
provides services through a SaaS platform to other technology companies
where they can track their faults and performance issues in
their software. And so, you know, we all sit there
(04:37):
and kind of cuss out our computers from time to
time when stuff just doesn't work. It's a fairly universal
problem and it's what Raygun helps with. So you can
think of Raygun as being like a black box flight recorder,
but the software code. So when something blows up, it
kind of gets the last message out to Sally, like, well,
this is what was going on at the time it
didn't work, and that is really helpful to the creators
(04:59):
of software, and so part to your question, you know,
why would kiwis not maybe know so much about us
as we're ninety three percent the export based, so most
of our money is we're pulling it in from overseas
and we're spending it here in Courtney Place, right, And
so we've got everything from small customers, new startups through
to Fortune fifty companies. So we've tracked for example, Domino's
(05:22):
Pizza Worldwide uses us, so if you're ever ordering a
Domino's pizza through the app whatever, and something goes wrong,
we collect the data, we helped them improve it. We
had Hbo originally as a customer and we tracked I
think it was eighty seven million concurrent viewers of the
finale A Game of Throne, So eighty seven million people
on Earth were sending data through our platform about how
(05:42):
long were they waiting for buffering? Was there any issues?
And so we operated a tremendous scale based just out
of here in Wellington. But that's our core business. That
was what we've been building for the last year ten years.
How did you get traction in the US?
Speaker 1 (05:57):
Initially that led to the Dominoes and HBO signing up
what was the breakthrough.
Speaker 2 (06:02):
Well, we've been selling products globally already under the brand
called Mindescape before that. That I started when I was
twenty three years old with a couple of other guys,
Jeremy Boyd and Andrew Peaters. They were the older, mature
people that I started the company with, and we sold
to developer tools there and then we built the raygun
product originally for Mindescape. It was our first subscription product.
(06:25):
I'd really admired what Rod had done with zero and
I thought, you know what, starting the month not at
zero is a good idea, and so we built a
subscription product around it and very quickly dominated all our
other products in terms of revenue. But simultaneously, because those
products are sold into software development teams, we knew from
(06:46):
literally day zero that we had to market this globally.
So you've got this thriving business. About twenty five people
involved in Reygun. Yeah, yeah, so have we. And this
is off a little bit at the back of the
AI story. We were actually pushing closer to sort of
sixty sixty five people a few years ago, and not
for you know, we haven't conducted rounds of layoffs or
(07:07):
anything like that, but when the AI stuff came along,
I kind of decided, well, if people were if people
were to leave, could we have a look at what
was necessary with technology to improve our output. I think
a lot of people kind of think that AI coming
along means they'll just do the AI version of their
current job. I'm increasingly not thinking that that is true.
(07:27):
I actually think it's an opportunity for people to really
think about what they like to do, and that suddenly
the how it's done is becoming a little less important
than it used to be. I'm not saying it's not
I'm not saying, you know that a person who, for example,
is never coded, is suddenly building operating systems, not at all.
But an example would be like these video creation tools
(07:49):
that have come out, and people are doing some amazing things.
And what I've noticed is it's not really the videographers
that are suddenly going, hey, I want to do this,
but it's it's seeing people that are genuinely quite caught
at their more creative that maybe it's like, maybe we
should have fifty times the number of directors now who
have creative talent that aren't blocked by needing an army
(08:09):
of people to deliver the output. Maybe there's more opportunities
for those types of roles to form. So anyway, backing
it up, so we let the head count generally sort
of drift a little down. We haven't seen any significant
slowdowns and productivity off the back of that. It wasn't
financially motivated, but it's becoming an attractor to see an
organization that kind of gets that this change is occurring,
(08:32):
and people's thinking, hey, well, at least go there. I'm
enabled to learn about how to move forward in this.
And I think that's one thing that is going to
catch a few organizations off guard if they kind of
continue to maybe play a wait and see approach, is
they're going to the people who don't really want to see. Yeah.
Speaker 1 (08:51):
Look, this is a trend we're seeing, particularly in Silicon Valley.
There's a lot of talk about this, about the fact
that you might have a company that can do one
hundred million dollars in revenue, but it does need more
than twenty people. Anymore, the expectation was as you scale,
you got more VC money or whatever you added, your headcount.
Speaker 2 (09:07):
Grew and grew and grew. It's not necessarily the case anymore. Yeah, well,
I think, and this is probably just me, but I
thought it was outrageous how much through Zerve all of
the product companies started to scale like they were services companies.
Like I totally understand if I'm building people out by
the hour, I need as many people as I can.
If I'm building a product that should scale out, I
should not need to keep scaling headcount as I scale
(09:29):
the product. The zero cost a duplication of software, that
is the benefit there. So I've always been a pretty
big proponent of smaller teams of highly capable people, very
empowered rather than giant factories. And in fact, you know,
probably a bit of a negative here, but I've been
saying this for about eighteen months now and I think
it's coming true, which is, if you want to avoid
(09:50):
the real risk of the downside of AI, do not
work in a company with tens of thousands of employees.
You know, you are by definition not wildly impactful, and
the ROI calculation is not going to be in your
favor when people are sitting there doing management by spreadsheets,
going oh, hang on a thousand people here, you know,
like that's some real money. I don't say that callously,
(10:12):
like I think it sucks, but I think that's what's
going to play out, and everybody should be sitting there
thinking about what's playing out.
Speaker 1 (10:18):
Well, that's playing out in the big tech companies over
the last two to three years. You know, we're seeing
four thousand people laid off in one go because the
spreadsheet being countered looking at that and going this is ridiculous,
as too many people in the books, and this is.
Speaker 2 (10:34):
Where you run the risk of zero sum game mentalities, right,
So there'll be a lot of people who freak out
kind of going, oh my god, the jobs are going away,
and it's like, well, no, the opportunity to create new
things are turning up, and you should be trying to
find what those are. So, you know, we were chatting
just beforehand about that Jeff Bezos quote where when you're
(10:54):
in a time of a lot of change, Jeff says
sort of the question isn't what's going to change, The
question is what's not going to change. And he gives
the example of Amazon where he's like, well, nobody's going
to turn around and tell me to ship things slower,
to increase the prices and all this. And that's where
I kind of look at this and I think, well,
at the end of the day, consumers are going to
continue to want to consume, you know, rightly or wrongly.
(11:16):
Human beings want more and more and more, you know.
And so you might sort of see an organization say
reducing head count but producing more. I don't think that
means that there isn't more organization starting up to also
produce things, you know. And so it really comes down
to that mindset. And I do worry a little bit
that the Kiwi mindset is too zero s. If somebody
(11:39):
else wins, there must be a loser. Therefore, you know,
we should penalize the winner. And it's like, no, you know,
we produce enough food to feed forty million people from
New Zealand. Maybe we could look at the next category
down and say, how did that actually served you know,
five x the number of people that will bring in
more money. That will be all that lifts the country
up is to bring more money into the country. And
(12:00):
you know what really bugs me about this, I've been
talking about this for years and now it sounds like
I'm talking Donald Trump's book, you know. I like, hey,
we shouldn't have a trade deficit. It's like when New
Zealand really shouldn't have a trade deficit, and it's had
one for nine out of the last ten years. Right,
we are progressively getting poorer and therefore we're asking a
lot of questions now, which are like, well, who can
(12:21):
we take the money off then if we can't figure
out how to earn it, And it's like no, no, no, no, guess,
let's work out how we can become hyper productive and
sell a whole lot more stuff to the world. That's
going to lift all the boats. You're going to increase
your tax take, you're going to increase people's quality of living.
All of these things come from business. They all come
from the private sector, and ideally in an exporting situation.
Speaker 1 (12:42):
Yeah, and look, we've got a government going for growth,
so that the rhetoric of the political rhetoric is in
that camp. You know, there's more momentum now, but our
execution has not been great in our trek record of this.
We have great companies who scale and employ people. You know,
take Sect. One hundred thousand dollars plus salaries. It's just
(13:03):
how do we do more of it? I guess is
the question.
Speaker 2 (13:05):
Yeah, and look there's lots of ways they could you know,
there's plenty of things we could do around improving our
R and D tax settings, for example. And the problem
is is that the tighter things become, you know, the
less options there are available. And I don't think many
Kiwis realize how tight they've already become. And that was why,
you know, part of the rhetoric at our auto Hive
(13:29):
launch is just sort of say, look, we can all
set around complaining and look, I've complained more than most
you know about about Winger's the government or the local
government going to step in and do something. But the
reality is there's actually society that does all of the work,
you know, and that includes the people in government. I'm
not sort of us them in that regard, but it's like,
(13:52):
why do I worry that too many people have since
at least since Az a kid has sort of shifted
their mentality here that they leave home and the government
becomes their parent, and it's like, what the hell are
you doing about? This is your shot to go and
make your impact in the world, you know, run out
there and do something amazing. Stop whinging about. I mean,
(14:13):
I have that issue when I see people saying there's
no jobs and I'm like, well, I've had exactly zero
people come and knock on my door saying, hey, can
I know your lawn's going to paint your shared Like,
I'm like, you know, where is the agency here? You know,
it's just no, no, the government should pay me more
money while I'm not working. Yeah, I'm not so sure
of that. I think we want and I think that
pioneer spirit is actually was the Kiwi spirit. I'm not
(14:36):
saying that to hammer people. I'm just sort okay, let's
bring that back. That was a much more fun attitude
to sort of look at the world through. Yeah.
Speaker 1 (14:44):
Yeah, So the other night you related to quite an
interesting story about early twenty twenty three. I think it
was the rise of generative AI. Chat ChiPT had come out.
The previous novembers would have wowed everyone, and you're sort
of planning the future of Reaguan and sitting down and
this time around was a bit different. You're like, I
(15:04):
don't really know where this is going. We really need
to think about what the impact. You said, it was
an electrifying moment playing with all this new generative AI
stuff and then thinking about how do we apply this
to Reagan and then thinking is there a business here
helping small and medium sized businesses use AI. Tell us
about that sort of that moment, that light bulb moment,
and what it meant for you as as an entrepreneur
(15:26):
and a business.
Speaker 2 (15:28):
To set the scene a little, I was in my
early teenage years when the Internet sort of came along,
and I was a huge nerd, so I loved playing
with that, you know, and I was cognizant at the
time that I was probably too young to sort of
make a go of building a business on the internet,
if you will, at that point. Now, admittedly I did
(15:49):
start building a business on the Internet when I was
twenty three, but in my teenage years, I didn't really
feel like I was in the position to really exploit
what the Internet was enabling. I spent a lot of
time reading about Microsoft and all the tech giants at
the time. You know, There's a great book called The
Microsoft Way that I enjoyed, and they talked a lot
(16:10):
about how those organizations were trying to grapple with this
shift in technology and so when this when the Internet,
so when the AI stuff sort of kicked off, this
to me, I sort of immediately recognized all the same
feelings and I think, you know, this probably applies to everybody,
but the older you get, you know, the more you
(16:30):
realize that your first inclinational gut feeling was absolutely correct,
you know. And that was why I was like, I
cannot put down thinking about this. This this is going
to be a bigger revolution than the Internet. It's going
to be bigger than computing. It's going to be the
biggest wave that mankind has ever seen. And if I
go back, I talked to the team. As you said,
(16:50):
I kind of went in and I didn't have a
clear vision for the year. I just knew that this
thing was colossal. I hadn't had enough time to play
with it. I hadn't formed strong enough opinions, And so
we spent the time talking about the nineties and what
did we see from the Bellgates Internet tied awave memo.
A few people realize this, but like Jeff Bezos was
(17:11):
working at a hedge fund before he started Amazon, and
the story goes that he's researching different industries and he
sees that the Internet is growing at twenty four hundred
percent per annum, and he just looks at this as
like nothing grows this fast. You know, something is going
on here, and we're seeing that, but potentially in order
of magnitude even bigger, you know, and adoption rates right now.
(17:32):
So I had this option where I was like, hey,
I've got this successful business. It's doing great stuff. That's
really cool. I see this giant thing happening. I definitely
need to apply some of that to what I'm doing today.
But also I didn't want to be completely handcuffed to
an old way of doing things. And so hence auto
Hive both as a separate brand with the intention later
(17:54):
of spinning it out as its own company. So it's
not pivoting Reagun Raguan just happens to be birthing this company.
And as you say, so we built. We sort of
had that time in early twenty twenty three. I had
the meeting with the whole team and was sort of saying, hey,
this is what's going on. We decided we then needed
to learn a lot more. So we had our AI
(18:16):
week in May twenty twenty three where we stopped all
work except for doing customer support. You know, I don't
want to leave a customer in the lurch, And in
the first four hours we got everybody to do our
engineering onboarding exam. So when we're hiring an engineer, you know,
we give them a test. And at the risk of
sounding a bit arrogant here, we have really really good
(18:37):
engineers at Reygun. Like I say, we scouted at eighty
seven million concurrent users for HBO. You know, I'm like,
look at trade meet, don't have to have everyone in
the country visit to just hit five point five million.
That was only one of our customers. So we know
how to scale, we know how to build international, and
so I wanted to so we built a range of
(18:59):
age to help us in our own business. These are
things like improving our AdWords, analyzing and bound trials, anything
we could kind of look at and think, what is
some toil that happens in our company that we could
take away, And so we built these agents. And the
key learning I sort of had was like, these are
amazing they have had they have improved our business, they
(19:21):
have allowed us to make better decisions. This is great,
but they're also really hard to build, and that you
needed this crack team of engineers to put this together.
And so that was exactly what you said earlier. Was
sort of the epiphany was like, hang on a minute,
if this technology is supposed to be running in every
single business out there and you need to crack squad
of software engineers to get any value out of it today,
(19:44):
that's not going to work. And so we set about saying,
how could we build a platform or product for the
everyman basically or every woman to say, you know, how
does the local accountant or the florist or the retailer
like take advantage technology. Now we're all kind of going
into chat GBT and asking questions and getting stuff, but
(20:06):
at the risk of over laboring the analogies. I think
of that back to the Internet is like when you
got email and you're like, you know what. Email was great.
You didn't have to go and spend twenty five cents
to send a message, and it didn't take three to
five days to get there, you know. So we suddenly
had free mail and that was kind of the hook
for the Internet, but it wasn't actually the major value
of the Internet. And I think with AI, what we've
(20:28):
seen is that the chat GPT is like the hook,
but the real value is coming in these agentic systems.
So appreciate the listeners who have probably gone this far
and maybe wondering what the hell and AI agent even
maybe we should explain it. Yeah, but yeah, let's talk
to that so and you know one of the things,
and like, if you're listening to this and it still
doesn't make sense, please send me an email at Janie
(20:50):
Trusk at autohive dot com and I'll try it again,
because I do value being able to take technical concepts
and try to make them approachable. But when you were
working with say chat GPT, and you ask it a question,
you might say, you know, what was the best test
score in the rugby last year, and it gives you
an answer. That's cool. You might ask it something that's
happened today and you'll see it go and do a
(21:11):
Google search for example. That is what's called tool use.
So imagine the AI has been given a hammer, but
the hammer is called Google search, and so it kind
of goes, well, I've got this in my tool belt,
should I need it, And you've asked me for something
that's you know, clearly happened only today, so I'm probably
going to need to go get some more information, so
it'll use that. Tool agents are really, in my opinion,
(21:36):
are just the tool using AIS. And so the question
then becomes how can I give it some extra tools
that would make sense to do more? So in our case,
for example, we use HubSpot for our help desk and marketing. Well,
let's give the LEM a tool that lets ask questions
of HubSpot and our help desk, and maybe it could
even send messages through it, you know, those sorts of things.
(21:58):
And so then we can have an agent that says, well,
I'm watching the help desk and I saw that our
new ticket came through and I read it, and you
know what, I actually have the tool to put the
draft reply on there, so that when you know, JD
or somebody on the team has turned up to work,
they can just review that it makes sense and clicked scent.
And so it's allowing these the AI to sort of
(22:19):
quote unquote break out of the box. How does it
start to interact with the world around us, even if
for now it's software based interaction. And so that's where
we have these examples that we you know, we had
a demonstration I think on the night playing in a
video where I made a simple agent didn't take more
than a handful of minutes, where I said, you're going
to help me find properties and I said, we're in
(22:42):
New Zealand, you know, consider local councils, all this sort
of stuff. You just put that text in the prompt right,
which is the main instruction. You write that in English.
And I gave it the ability to search the web
and read web pages. That was all I did. And
now when I say, hey, I'm looking for a property
I wanted to spend you know, so one point two
million dollars in Mount Verk it needs to be in
(23:04):
you know, the school range of this thing. And you
watch this agent go off and spend the next sort
of ten to fifteen minutes browsing real estate, dot code
and Z trade me cross. It builds up a list
of like properties that's going to suggest. Then it goes
and checks the council websites and it gives you the
questions you might want to ask the real estate agent.
And this is just one simple example you know that
(23:26):
you can put together and then you start looking at
all these for business. So like I say the question,
I always always ask people, and I encourage the listeners
to think about this is just what is a task
that takes you one hour or more every week or month?
You just fre again hate everybody has a few and
(23:47):
voicing exactly. You have these things where you kind of go, gosh,
if somebody took that away, you know, I'd go and
focus in these areas. And that's that's kind of a
little bit what I think about when it comes back
to that disruption thing. It's like you're really ina blling
people to focus in the areas they actually want to
put the effort into, you know, and play to those
strengths and take away the toil work that's not actually
(24:08):
much fun for anybody.
Speaker 1 (24:10):
It's those time consuming tasks like invoicing or customer care
and automating that. But you've built a platform sitting behind
at is the large language models. We're using it in
a consumer sense, the open ayes models, presumably Claude and others.
So you're using the best large language model to apply
(24:31):
to a particular task, and you've built a beautiful interface
that makes it really simple to build your own agents.
Speaker 2 (24:38):
Yeah, one hundred percent. So first off, it's I'm sure
the folks that are paying a bit of attention today,
I see this. You know this week Claude will be
the best. Next week CHATBT the week after that, you know,
Elon might finally ship something you know, all of these
things keep playing out, and so we want to sort
of say, hey, it's kind of it's somewhat irrelevant. You
(24:59):
should be to use whatever the best thing is the
time through our platform. You're right, The entire UI and
everything we're working on is to try and make something
that is not for nerds. We want this to be
for people who say I want to jump in and
as you say, maybe I want to automate a bit
of invoicing. I want to be able to like toggle
on the zero integration, you know, and say, hey, give
(25:21):
me a more readable version of my pen now, or
just describe what the biggest risks are or things like that.
You want to make people so they're empowered to do stuff.
One of the things we're trying really hard to do
as well is avoid all of the acronyms. Like you know,
I've been in the tech industry for a long enough
time that that's usually the first place we lose people. API. Yeah,
(25:42):
my API talking to the RAG system so that you know,
my AI can You're like, what the hell are you
talking about? You know, And so often the tech industry,
I think rightly gets a bit of an eye roll
because it almost feels like we set out to make
things more complicated than they should be. So yeah, we're
trying to make something where it's hey, leave the really
nerdy stuff on the back end to us. We'll figure
(26:03):
that out. I probably wouldn't talk about this on an
international podcast like mono Hive will be an international product,
but I want to have that impact on our local
businesses in New Zealand. We know we have a productivity issue.
We know this technology is coming, there is now something
home growing for it, and so we get a huge
amount of value from people just feeding back one are
the bits that suck. You know, it might sound really
(26:25):
weird to say, but it's like when you launch version one,
it's the worst state it's ever going to be in
for the consumer, right, so we want to learn as
fast as we can. You mentioned HubSpot.
Speaker 1 (26:34):
You've got relationships with a number of box Gmail for instance,
So this is very much you know, it relies on
you getting that often sensitive but high quality company data
that's sitting in various data repositories. We've heard so much
from the big tech companies that have gone all in
on agents salesforce. In particular, Mike Benioff is betting the
(26:57):
company basically on agents, Microsoft, SA, all of them, and
their whole thing is bring your data to us, keep
it really clean on our platform, and it will allow
you to do all of this sort of stuff. How
do you approach that because you haven't got hold of
everyone's data in one place. Are we back to the
so called APIs to get that data into autohive?
Speaker 2 (27:19):
Yeah, so let's break that question down. So firstly, all
the big guys want to be the hub in the wheel,
so they want all the data to bang on, you know.
And I always think if Mark Benioff really believed in
what he was doing, he wouldn't be selling three million
dollars with a stock every day, which he does. So
you know, we've been a Salesforce customer, and to be honest,
(27:40):
I think we're going to see some bad behavior off
the back of AI. For example, Salesforce have recently said
to some companies that they will prevent them from integrating
into Slack because they don't want AI companies to be
able to get that data. I think I own the
damn data, but what my team are talking to talking
about in Slack, and so I think we're going to
actually start to see some of these particularly big companies
(28:02):
start becoming quite bad actors about who actually owns company data.
And I don't think Salesforce is actually doing a very
good job of demonstrating good behavior. Now that's a very
Salesforce specific answer for a moment, but ultimately I think
that as much as anybody would love to sort of
own a business where every party puts one hundred percent
(28:23):
of their work on your system, it is it is
naive and foolish. I think the reality is we actually
work in a mixed world of various systems. You know,
I'm not going to suddenly move my accounting of zero
to something that Salesforce makes just to improve an AI.
I want the AI to be able to talk to
zero and do a great job. And then what we're
seeing out of a lot of the foundation or the
(28:45):
big call them the big guys if you want. And
this is a big differentiation from autohyper auto Hive should
be the glue between all of the systems that you've got.
It should be the interface that you go and use
to work with. Primarily, it should be nice and joyful
to use. Matter that you're pulling data out of Google Drive,
out of zero out of Gerro, out of HubSpot, and
we bring all of that together, we take care of
(29:07):
the messiness of how to do that behind the scenes
for people. And I think when you look at some
of these eye watering valuations on these AI companies, look,
maybe I'm just under and I already think I'm estimating
a pretty big size and scale of AI, but maybe
I'm underestimating. But some of those valuations seem to be
(29:28):
predicated on the idea that those AI companies completely own AI.
And I'd be remiss if I didn't sort of highlight this.
You know, we have some folks even in our business,
they definitely believe as an AI, but some of the
first call bs on the hype and I'm absolutely one
of those people too. We've talked a bit about the
(29:49):
potential around job losses and things like that, but the
reality is today it's about automating a handful of hours
every week in toil right. Technology very well may improve
over the coming years to make that more impactful and
more cavble. I have no doubt of that, But today
it's about save yourself a little bit of pain overall
(30:10):
in there. I don't think people want to be starting
that journey with I have to pile it all on
to one megatech.
Speaker 1 (30:17):
So it's been out there, you know, only launched it
maybe a couple of weeks ago. But what's been the feedback,
like in some of the use cases that are starting
to bubble up with auto high.
Speaker 2 (30:26):
Yeah, so we actually launched it silently last Tuesday, and
then Wednesday was there, let's start making some noise on it.
So it's not actually been out a week yet, so
we're pushing four hundred odd folks who have now come
in and tried it out, which is really cool. We
can see folks have made a range of agents. I
(30:47):
will say we do full This sort of touches on
an earlier question, but there's full encryption across all of
the conversations. We can't actually read, you know, I can't
sort of sit there and go, oh, Peter, I asked
this question. That's all lockdown, and we're very mindful about
the trust and security side of things. But we can
see people making agents that they are sort of setting
(31:08):
up these different tools to run things. And I have
had a few people citing the examples like the property
one that I made, which is how I know exactly
what that was doing in there. The main piece of
addressable feedback that we'll be tackling over the next like
I said a couple of months, is that marketplace piece.
So it's great to come in and you can use
(31:30):
all of these l elems. You can use Chantypete, Claude,
Gemini all in here, and we actually have a you know,
it's free to start. You know, we have ten dollars
worth of credits on there, so you can just use
it for free to begin with. Then we have a
paid tire where we go to eighty dollars a month,
but this is not per user. So if you've got
a company of one hundred people, pay eighty bucks and
(31:51):
put them in there. Now, there is some usage pricing
that can go on top of that later on. So
if you do have a few people that are really
hardcore and they're using a lot, just be the price
of the tokens plus a small margin on there. Now,
why this is useful is because, and I can say
this even across the tech industry for those that are
listening that maybe aren't in tech companies, the adoption of
(32:14):
AI is very sort of in disparate groups right now.
So even in software companies, you're sort of seeing twenty
maybe twenty percent of people are really leaning in and
discovering what's possible. There's about fifty percent of people are
sort of, you know, maybe asking a handful of questions
each week, you know, and then twenty five percent of
(32:34):
people who are just not at all engaged. The problem
with that model is, obviously, if you're the CEO, you
want to get your team up to speed. You want
to roll out all the tools. If I was to
go and get Gemini, Claude and chatchby t for a
single team member, that would add up to sixty US
dollars for that one team member right now, if that
team member is in the group that isn't really using
(32:56):
it a lot, I'm going to start getting frustrated that
I'm paying for something that's not being you. And so
that's why we've tried to make the model one where
it's like, yes, there's the starting price, but that gives
you enough credits to sort of not hit the wall
too quickly. But then you're only going to scale it
based on if you're getting the utilization. And that's where
I think ultimately we want to be aligned with the
(33:16):
customers is to say it's in our interests and your
interests that we're delivering creating agents that are delivering real
business value. Because if I can get ten thousand dollars
worth of ROI out of two hundred dollars worth of spend,
I will do that deal every day of the week.
Speaker 1 (33:32):
Yeah, and for a small so you've got a ten
person business or you know, twenty or thirty person business,
eighty bucks a months a month. If that's spread across
various agents, because you can set up multiple agents and
I see you know, they even have their own avatars.
So increasingly we're going to be working along these different agents.
(33:53):
You'll have the agent for invoicing, you'll have the one
for customer service, and in auto hive you'll see the
progress where what they're doing, what they're reporting, where they're at.
Speaker 2 (34:02):
Yes, one hundred percent. So for those of you that
are listening, imagine a user interface that looks a little
bit like Slack or teams, but we say half of
the users are actually smart agents. So you might have
like a governance one that understands your shareholder agreement. You
might have I've made one that advises me on my
personal stock portfolio. You know, it can read that data
(34:24):
and it just sort of says, hey, you know, you
might want to rebalance this way, or here's the earnings
reports coming up, those sorts of things, and so what's
kind of fun about it is, don't think about bringing
autohib into your company, like, okay, cool, we need to
mentally spend you know, one hundred thousand dollars this big
bang deployment. We need all these agencies. He no, think
(34:44):
of it as like you could go and sign up
as the CE even you know, you should be able
to make a couple of agents to do something simple
and then you can invite your team in for free.
They don't cost anything more. It feels a bit more
like an infinite box of Lego, you know, where it's like, cool,
build something, just start to explore, you know, and then
you'll start realizing, oh, I could build even fancier things
(35:06):
and fancier things. And we found this even in building it. Right,
So we would have a discussion internally where somebody said, oh,
how could we build an agent that does X, Y Z,
And I was like, well, I'd probably do it. ABC
one of the younger team members was like, actually, we
could just have the agent talk to another agent and
do this, you know, And that's the power of the
(35:27):
creative thinking. Like a lot of us that are a
little bit older kind of have already got an idea
of how to work, and these things are quite a shift.
And so that's why even with an auto I've mentioned
the easy to use you I trying to make it finally,
sacronyms and all of that.
Speaker 1 (35:42):
I guess the big promise from the Benioffs and that
is we've moved from this information generating ability with generative AI,
which is incredibly useful, through to doing stuff on your behalf.
And I remember twenty eighteen sitting at Google iOS conference
where they debut Google Duplex, which gone nowhere. But they
made a call or this AI agent made a call
(36:04):
to a salon and had a conversation. This was an
AI bot having a conversation with the salon owner to
book an appointment. So the point where we get to
aim making decisions on our behalf and adjusting if the
salon owner goes, we're booked out this week, you know
what about next week? Going and consulting your calendar to
see if you have a slot next week. How far
(36:26):
away are we from that sort of stuff, we're pretty
much there there.
Speaker 2 (36:30):
Yeah. I mean the calendar example, just on the auto
hive side, I played with this. I just attached my
Google calendar and I started asking it questions like what,
you know, how would I optimize my week and things
like that. So the pieces are there, They're all on
the table. I often. That's why there is just fundamentally like,
the amount of opportunity in AI right now is just insane.
(36:53):
Any direction you look into is just a blue ocean
of opportunity right now. That I used to think when
I was building reagun, Gosh, wouldn't it be nice to
be in the blue ocean rather than like competitive field?
And now I kind of take the viewing like, boy,
it's quite disabling to have infinite opportunities in all directions
(37:14):
and how you kind of keep that focus, Which is
why we keep coming back to that SMBs, which by
New Zealand standards pretty much everybody you know, and ease
of use, you know, that's what we're going for. But
the ability to glue these things together to create these
complete experiences is there absolutely today I have to use
(37:34):
the iPod. It's a good example of that, which is
The original story of that was that one of the
hard drive makers I forget which one, had actually told
Apple that they were building this very very tiny hard
drive and they had no idea what to use it for.
(37:55):
And it was Steve Jobs and Johnny I've that kind
of sat down and we're like, well, hang on, if
we can put this much memory into its device of
this size, you know, we could do this, this and
two hundred songs. Yeah. But this is the thing I
always think is most interesting about a lot of innovation
is how frequently you kind of look back and you
realize all the pieces were on the table. It just
took a person working out the right way to put
(38:17):
them together that they make these phenomenal products. And I
feel right now like there is a tremendous amount of
pieces on the table, and we're going to spend years
building the ability to leverage it. Yeah.
Speaker 1 (38:29):
So what's the plan for the you know, the remainder
of twenty twenty five. You're really building this with the
local focus first, but clearly trying to prove some use cases,
get some feedback from customers, very with a view to
very quickly taking it international.
Speaker 2 (38:46):
Yeah. Absolutely, So there's nothing that prevents international people signing
up today. And because we do have the background in
Reagun like Reagun does actually have people outside in New
Zealand too, so we've already got you know, like you know,
we're already set up for account in the US and
we've already got the tax stuff and security and all
that sort of stuff is already in place, which is
good between now and the end of the year. The
(39:10):
main one is that marketplace capability. And I don't want
people to think of the marketplace is purely like I
have to go spend money. As you say, there could
be a lot of free agents on there, but it's
that ability for people to almost pick from a menu
rather than have to make something. You know, I really
want to dial that up to eleven. So those use cases.
We do have more than a thousand use cases listed
on the site, but just generally, how do we lower
(39:31):
that barrier to entry. So between now and Christmas you'll
get to a point where you might search for say
you might kind of go agent to maybe do zero invoicing.
You know, you want to have that come up and say,
well here's the page on the autohib site with you know,
there might be three of these agents. They should have reviews.
Click on that. Oh yeah, I like to look at
that one, but click get sign them with Google and
(39:52):
be dropped straight into talking to that agent you know,
able to assist you get people into the ecosystem. There.
One observation is just that we well have to put
a lot of work and this is a good thing actually,
and to really leveling up people's understanding of how to
even prompt these tools, there is definitely a real skill
(40:15):
in that. I think again, just as a tip maybe
for the people listening is I've often found that a
mistake people make is that they give too little context,
so they only maybe give one or two sentences, which
works really well if you say, to say, chat GBT,
you know who won the presidency in nineteen eighty four,
that's fine, But if you want to actually do something
(40:36):
somewhat complex, you've got to think of the prompt as
sort of setting the parameters in which the AI should
adhere to and then be open to accepting all possible outcomes. So,
you know, this is where we get the crazy stories
from the movies where it's like, you know, is it
two thousand and one Space Odyssey, where it kills everybody
(40:58):
on the ship. Now, you just said you needed the
people over here, and so they had to be alive, right,
And it's like that's the bit that trips people up
and it sounds like we're being insulting to AI. But actually,
if you think about it as just a general intelligence,
if you didn't say these things, you know, like, how
(41:19):
would it know? And so oftentimes when I see people struggling,
it's either putting in two little context or that're trying
to micromanager, like they literally want it to spit out,
like they know to the letter what they want in
the outcome to be. And it's like, don't do that either,
like you wouldn't you wouldn't expect that when you're managing,
say a junior team member either it's not going to
(41:39):
be a good way to work with them.
Speaker 1 (41:41):
Yeah, And there's just been so many reports. Microsoft just
published one about the productivity potential of AI, you know,
so I've covered so many of these reports, but it
just seems to be a gap between what's possible in
the short term in the next five to ten years,
and are ability to exploit that. And it really comes
(42:02):
down to digital skills and people having the confidence to
actually get hands on with these tools.
Speaker 2 (42:07):
We've got a lot of work to do there. Yeah,
at a fundamental level, totally agree with that. I do
think to your point, there's a lot of reports and
a lot of noise coming out of these big companies.
I think they are over hyping things a bit. Simultaneously,
I think they are also fair mungering way too much.
They're creating them. I mean I saw a thing the
(42:27):
other day where one of them was like eighty five
percent of the time the anthropic model was blackmailing people.
I'm like, you would have had to go out of
your way to prompt that thing to do this, and
you're trying to scare the bejeebus out of people. You
know this, This is a bit crazy. So, as a
few people sort of highlight, a lot of those big
(42:47):
companies are really pushing to try and create a regulatory
capture environment where only a handful of them are allowed
to operate. Yes, and it's it's not cool.
Speaker 1 (42:56):
Are you bootstrapping this out of your revenues from Reagan?
Speaker 2 (43:00):
Yeah? So reagun is a fairly well established business. So
at the moment it's being funded through reagun. And once
we've driven up revenue. Like I say, the intention will
actually be to spin it out as a standalone company
rather than keep it as a business unit. Reagun itself,
like I said, is doing fine. It's still very valuable tool.
(43:21):
We're going to continue to add AI capability so that
we already have some in there overall. So yeah, that's
the that's the approach we're taking at the moment.
Speaker 1 (43:31):
Yeah, so it's really about getting those use cases for
auto hive. People come back going Wow, this is saving
me a lot of time and energy, and then applying
that hopefully to others.
Speaker 2 (43:42):
Yeah, this is soon one hundred percent. And your point
earlier about you know, the over hyping from the big guys.
That's why I say start rather than sort of having
this existential crisis of like if I bring this in,
am I implying that I don't need stuff, and like, no,
you're not. You're probably going to automate some of the
more boring things that happen in your business, and you're
(44:04):
probably going to find the wow factor is really just
to be able to say that some of that work
has been taken off your hands and there The other
thing I'd like to just address is I think we're
drawing to a close here, as you mentioned as well,
having these things do stuff on your behalf, and I
want to address that just because I'm sure there's some
people they're thinking, gosh, there is no way in heck
(44:28):
that I want this thing to do stuff on my behalf,
you know. So what we've done is we have what
is called human in the loop. And this is somewhat
common or becoming more common, but it's the idea that
you can choose the actions that need your explicit approval.
So the email example I have was I attach it
to my inbox and I could say what's my writing
(44:49):
style or give me some tips on improving of you know,
who's my most annoying person? Email? But I can choose
that to send an email will require that auto Hive
actually prompts me and says I'm going to send this yes, no?
Or do I want to interject and say maybe tweak
the message this way first and show it to me
so you have absolute control on what you're going to
(45:12):
allow it to do autonomously. That's how a lot of
the very cutting edge tools are operating. And my vision
for this and you asked about between now and the
end of the year, and I talked about the use
cases with the marketplace. We'll also have a mobile app
by the end of the year. My dream for this
is I want to wake up in the morning and
I want to have had these autohive agents have drafted
(45:33):
or my email replies. I want them to have read
my calendar and see that I have a board meeting
and I've got this meeting at this meeting, and I
want it to prepare all the work as far as
those agents can take it, but they won't necessarily have
completed it, so you know, it might have drafted, say
five email replies. I want to be able to get
out of bed. Maybe I'm still lying in bed, you know,
look at my phone. Yes, yes, no, no, yes, you know,
(45:54):
I'm like, you know what, before I've even got out
of bed, I've actually just kind of cleared a bunch
of this. Yeah, all's work. That's the goal, you know. Ultimately,
it's about trying to free people up to focus on
the stuff they actually enjoy doing. I don't. I don't
just like the people who send me emails, but running
emails is not my top most interesting task of the day, or.
Speaker 1 (46:16):
Spending time on in New Zealand's website, flicking through looking
for flights and all that sort of thing, if I
can get the agent to just come back to me
and go, this is what I found. Do you want
me to go ahead and book that for you? I
think a lot of people be comfortable with that.
Speaker 2 (46:28):
Yes, yeah, I was using that a fair bit when
I was traveling in the US, you know, like I
want to be in this area, I want to spend
this sort of money. Find me some options, you know,
just go and have a look.
Speaker 1 (46:37):
Yeah, great, Well, thanks JD for the rundown and all
of that and on agents and where they're going. Good
luck with auto hive and as you've said, you're happy
to help anyone who wants to see it one up.
So that's a great opportunity for anyone listening.
Speaker 2 (46:50):
Yeah, one hundred percent autohive dot com. It's free to
sign up. There's some free credits so on there as well,
and you can talk within our community and we're happy
to help. Ultimately, I just want to see you said,
I'm elevating on this and doing better with AI because
I think it can actually help us with some of
the challenges we face today. That's it.
Speaker 1 (47:13):
For this episode if the Business of Tech. A huge
thanks to JD Trask for sharing his vision and some
practical wisdom on building with AI, not just for tech giants,
but for every business ready to take the leap. And
if you are ready to do that and see what
AI agents can do for you, check out autohive. Sign
Up is free at autohive dot com. The credits are
(47:36):
on the house the first ten dollars worth anyway, and
JD and his team are eager to help you get started.
At businesses dot co dot enz. In the podcast section
you'll find my show notes with so many other things
that JD mentioned, for instance, the Bill Gates famous Internet
title Wave Memo, which is worth a read. Find them
in the podcast section at businessdes dot co dot anz.
(47:59):
Don't let the future automate you out, that's the real message.
Get ahead and automate the future yourself. Next week, a
rising star of our space sector attempting to solve a
really big problem facing the companies that are launching ever
more satellites into orbit around the world. That's next Thursday.
(48:19):
I'm Peter Griffin. I'll catch you then,
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.