What happens when artificial intelligence evolves beyond simply responding to prompts? Welcome to the world of AI agents—autonomous systems that observe, reason, and act with purpose to accomplish complex tasks without constant human oversight.
In this enlightening conversation with Vic Rogers from Sustainable Growth Creative, we dive deep into what makes AI agents different from traditional large language models and how they're already transforming industries. "AI agents are software that perceives, reasons and acts towards a goal," explains Vic, highlighting how these systems leverage LLMs as their "brain" while adding critical capabilities like memory, planning, and specialized tools.
The real-world applications we explore are both exciting and potentially disruptive. Customer service call centers are incorporating agent technology to handle inquiries autonomously. Marketing departments use agents to create content and optimize SEO strategies. Perhaps most revolutionary is software development, where "weeks turn into days" when developers pair with AI agents, potentially requiring fewer human coders while dramatically increasing productivity.
We also examine the fascinating realm of multi-agent systems, where independent AI agents interact with each other. Stanford's Smallville project demonstrated how 25 separate agents developed social behaviors like organizing birthday celebrations without explicit programming. In the business world, these multi-agent ecosystems are enabling everything from supply chain optimization to decentralized finance trading.
The distinction between assistants (which augment human capabilities) and fully autonomous agents becomes especially important as organizations consider implementation strategies. For regulated industries concerned with compliance, the human-in-the-loop approach of assistants provides necessary safeguards, while other sectors may benefit from the efficiency of fully autonomous solutions.
Looking ahead, personal AI agents could transform how we manage our daily lives. From scheduling and email management to financial monitoring and meal planning, statistics suggest a quarter of digital tasks could be handled by personal agents within a year. Companies like Rabbit and cognitive labs are already working to make this technology accessible to everyday users.
Connect with us at AI Ready RVA to stay informed about this rapidly evolving technology landscape, and learn how your organization can prepare for the agentic AI revolution that's not just coming—it's already here.
Want to join a community of AI learners and enthusiasts? AI Ready RVA is leading the conversation and is rapidly rising as a hub for AI in the Richmond Region. Become a member and support our AI literacy initiatives.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Welcome RVA to Inspire AI, where we spotlight
companies and individuals in theregion who are pioneering the
development and use ofartificial intelligence.
I'm Jason McGinty from AI ReadyRVA.
At AI Ready RVA, our mission isto cultivate AI literacy in the
greater Richmond region throughawareness, community engagement
(00:24):
, education and advocacy.
Today's episode is madepossible by Modern Ancients
driving innovation with purpose.
Modern Ancients uses AI andstrategic insight to help
businesses create lasting,positive change with their
unique journey consultingpractice.
(00:44):
Find out more about how yourbusiness can grow at
modernagentscom, and thanks toour listeners for tuning in
today.
If you or your company wouldlike to be featured in the
Inspire AI Richmond episode,please drop us a message.
Don't forget to like, share orfollow our content and stay up
(01:07):
to date on the latest events forAI Ready RVA.
All right, welcome back toInspire AI.
Today we have an openconversation with Vic Rogers
from Sustainable Growth Creative.
He's a regular on the show.
We've had him talk about hiscompany, his role in AI Ready
(01:28):
RVA as a board of directors andhis enthusiasm for AI agents,
which is the future ofartificial intelligence,
autonomous nature, and here weare to discuss some of our
recent findings and to create aspace where this is just an
(01:48):
approachable share out of ourlatest thinking, so we want to
talk a little bit about what'sthe latest in the technology,
not so much in a deep dive oftechnical components, but how
does this stuff work, where isit used, and we'll get into some
of that.
So welcome back and it's greatto have you on Vic.
Speaker 2 (02:10):
Oh yeah, man, Looking forward to it.
I think this is goodPre-homework before the event on
Tuesday, with the AI agentsession going on, so I'm looking
forward to getting a little bitstronger in this subject matter
.
Oh, yeah, tell us a little bitabout that, yeah, so, depending
on when this comes out, I'm notsure which cohort that's coming
(02:30):
under, but there's going to be,on April 29th, an AI agents what
they are and what they aren'tsession going to be ran by a
corporation called Rise andScaleai and it's going to be a
Dominion payroll from 530 to 7.
It's one of those things where,if you're interested in just
(02:51):
understanding what AI is, we gota heavyweight coming in, a
young lady by the name ofJessica Clark who's a senior
solutions architect at NVIDIA.
She's going to be doing a lotof education.
I think it just goes well.
This is just a pre-work beforethat, or, if you listen to it
afterward, you know same deal.
You know lock in with what theygot going on over there support
, support the sponsor, rise inscale, and you know if you can,
(03:23):
if you ever get a chance to talkto somebody from NVIDIA about
this stuff.
You know I think right nowthey're at a pace center when it
comes to like being ahead in amarket with it's, with this
information.
So I've been doing a littleresearch just to make sure when
I go in and this doesn't go allover my head good, yeah, well,
uh, dominion payroll.
Speaker 1 (03:32):
Tuesday evening, nvidia's architect, jessica
clark, be giving us a greatpresentation.
I don't have any guaranteesthat this episode is going to
come out before then, but getyour tickets.
Yes, sir.
All right, follow us onLinkedIn AI Ready RVA for the
latest news as well as events.
(03:54):
So let's get this conversationstarted, vic.
Let's talk about what AI agentsreally are.
What do you think they are, vic?
Speaker 2 (04:06):
I think it's software that perceives reasons and acts
towards a goal.
It's a little bit more thanwhat you see with the actual LLM
.
You know, when you think ofLLMs, you know they're taught in
their power from an externaldatabase.
You know a lot of these agents.
They're going to be rules-basedbots and they're going to have
(04:28):
scripts and those agents aregoing to be able to adapt in
real time, which is amazing,right?
So I personally am lookingforward to like this year being
the year of agentic AI.
You know that term just soundsfunny, but that's what the
Zeitgeist is.
You know, using agentic AI.
You know that term just soundsfunny, but that's what the
Zeitgeist is using agentic AI.
(04:48):
And you know, when I did myresearch, that's how I came up
with it.
What about yourself?
What are your thoughts?
Speaker 1 (04:54):
Yeah, I think agents are advanced autonomous systems,
right, like.
You have an LLM that is kind ofthe brain of the agent, and
then the agent is the programthat leverages the LLM and it'll
leverage other tools Maybe wecan talk about some of those
(05:15):
tools as well to autonomously,like you said, reason about the
environment.
So it observes and then itreasons and then it takes action
, right, like.
I think that's what the at theat the end of the day, the crux
of the of what an AI agent is,it's able to observe, reason and
then act upon that, and thethinking part of it right is is
(05:39):
really just about, like, givingit context.
So the LLM itself, theabundance of information in the
internet, right, can be used inthe agent's observations and
reasoning.
Right, it can think about whatkind of information it's
available to leverage and how itmight take what it's observed
(06:03):
and think a little bit morestrategically or tactically
about what its task is.
So there you go.
I think those are the types ofthings that jump out at me when
I think about it.
I think there's a lot ofapplications to AI agents.
You know you got your robotics,you've got your customer
(06:24):
service support centers and whatelse?
Where else might you see agentsthese days, vic?
Speaker 2 (06:31):
Customer care, 100%.
You know I think that callcenter work is going to be
replaced pretty soon with a lotof agents.
And I know, don't quote me onthis one I can't remember the
actual fast food company, but itmight've been Wendy's actually,
or I might be confusing Wendy'ssuper side note, like with
(06:54):
Palantir.
I know Palantir and Wendy's areworking one another and that's
kind of weird.
But just having, I think at onepoint that there was a goal to
try to use ai to take drop totake menu orders, you know, and
it didn't work out.
But like I think they'rethey're refining that a little
bit better.
But then also with likemarketing, like right now, you
(07:15):
know, um, you can have a oneperson news crew where you know
you can use the ai agents tocreate blog posts and then just
make sure the seo is where itneeds to be and then poof, you
know you're out there with somegood content.
I think the one that's probablythe scariest, especially for,
like time spent in the space,software development.
Like you know, if agents canstart coding, like you know,
(07:37):
it's kind of, like you know,tough for, like you know,
developers, coders, architects,who have been in the space for a
long time to have thistechnology come through because,
like, what's going to happen islike I don't know if I say one
in four or you know it's a nicejump, but like you won't
necessarily need to have thestaff to hire as many of these
architects and coders anymorebecause you could just have an
ai agent paired with a human todo the same amount of work 10
(08:00):
times faster.
Right um, weeks turn, turn intodays when you're looking into
that Finance too.
You know just a lot ofdifferent use cases and you know
, I would say you know picturinglike nonprofits, getting into
leveraging.
You know these agents withouthaving to hire, like data
science teams, you know.
So there's a lot of good and alot of displacement, but if used
(08:22):
properly in the rightindustries, the right use cases,
I could see a lot of positivehappening.
Speaker 1 (08:26):
Yeah, yeah, I think we'll have an opportunity to
talk about AI autonomous humanreplacement versus augmentation.
I think that we're on a path toaugment right now.
Full replacement isn't so muchin the short term for us right
now, because I think thatthere's still some concerns
(08:49):
about the reliability of itsoutputs Right, and nobody wants
to be put on the front page for,you know, last week releasing
an autonomous agent and thenthis week, you know, blowing up
their industry, right?
So I think there's there'sstill going going to be some
work to do before fullreplacement.
But, yes, augmentation, as wesay, is definitely very
(09:13):
approachable for companies andthey are considering that at
full speed.
So we've gotten to the pointwhere we know what an AI agent
is, you know, logically speaking, and we know how it's being
used in companies, industries.
What about the building blocksof AI agents?
(09:35):
Let's click on that one alittle bit.
What do you think are importantcomponents to creating an agent
and making it successfullyoperate?
Speaker 2 (09:48):
I guess a couple of things.
I think a good analogy I sawone time was that you know AI
agents are like Lego sets, likeLLMs are leveraged for thinking
and then tools for acting.
You know memory for learningand then planning for strategy.
So like that's a good kind oflike way to you know, think
about it.
Or even when you think aboutthe building blocks, you know
(10:09):
you got a brain, you got atoolbox, you got a diary, you
got a whiteboard and then youwire them all together and bow,
you got this beautiful AI toolright.
So just broken down and like alayman term, I think that's
that's the easiest way to kindof like fill it out when it
comes to like the buildingblocks of.
Speaker 1 (10:26):
Okay, there are some, definitely some tools out there
.
Let's let's take, for instance,an agent that can scrape any
website and help formulate anopinion of you know what a
company does, how your businesscan support it, that sort of
thing.
When I think about that usecase, this agent could be using
(10:48):
a, you know, browser type searchtool, a scraper, something that
will pull information fromspecific website or web pages
and then some sort ofsummarization capabilities with
the LLM, then use some sort ofRAG-like so retrieval,
augmentation, generation,RAG-like thought process to
(11:12):
review, based on your companyopportunities or service-level
opportunities for your companyto help with and possibly make a
pitch for.
So all of that is the types oftool sets that agents can use in
(11:51):
just that one circumstance.
I think that I want to talk alittle bit more about how it
does that.
So humans use memory to observeand reflect on things.
Agents do the same thing, right.
Agents do the same thing, right.
So in that case, how do youthink that the memory would
(12:12):
surface and be of support inthat structure?
What do you think about that,Vic?
Speaker 2 (12:19):
Vector DBs such as Pinecone and Prestigious can
keep past calls so the agentscan actually remember your
specific use cases, like if it'slike organizational policies
and things of that nature.
I think from a memorystandpoint, you know that's
(12:41):
ideal Persistent storage, justenabling agents to recall past
interactions and improve overtime and like build over time.
I think both of those are umexamples of how like memory is
going to be leveraged.
When it comes to using these umai agents and actually I was
gonna say in the future, butlike now, currently like there's
some tools out there that we'llprobably talk to eventually I
(13:02):
haven't seen a lot of like nocode easy to use ones yet I've
been playing trying to figureout some for a client of mine.
But um you, it comes to memory.
That's what I'm seeing some ofthe best speak to.
Speaker 1 (13:16):
The Pinecone one I've been using recently in building
a chatbot, just testing it outand seeing how scraping a
website and storing it into thevector database, which is
PinecCone right, how that works,being able to access that
(13:36):
memory, if you will, from thechatbot's interface, asking the
bot to recall what it learned byjust submitting some basic
questions about the website thatit ingested, see what it knows,
and so far it does reallyreally well.
So a little bit of explorationthere goes a long way, for sure.
Speaker 2 (14:01):
How long have you been doing that too?
That sounds kind of cool.
So you're already in there,kind of coding an AI agent.
Speaker 1 (14:08):
Well, actually I used Claude Claude 3.7, Sonnet, I
think it's what it's called andjust the free version if you, if
you know how to instruct it tocreate something like that's.
What I've been doing lately isjust testing out its
capabilities for producing anentire application like that.
(14:29):
I'm happy to give you a demoafter the podcast, but it's
pretty amazing to be able to seewhat it can produce.
And the problem with it is thatif you have it build something
that's extravagant and youactually get it to work right,
but you don't really know howit's working, then you can't
(14:53):
really update it right, youcan't maintain it.
You ask it to do something elseagain, to try to like adapt it
or create a you know, a newfeature or something like that,
and it'll go back through andrewrite the entire code, which
changes things right, and youhave no idea what it's changing.
So that becomes one of theblockers for it these days.
Speaker 2 (15:18):
But I'm happy to talk with you a little bit more
about that after the episodeDefinitely one of the benefits
of being a co-host getting coolstuff like that man.
So look forward to thatconversation.
Speaker 1 (15:30):
Yeah, definitely getting getting cool stuff like
that man.
So look forward to thatconversation.
Yeah, definitely, um, it was.
It was kyle.
Uh, johnson from our platforms,who you know shared with me his
insights into claude and how hewas using it and I was like I
need to go try that.
And so there you go.
Um, tell me what.
What are your thoughts on theagent versus assistant debate?
(15:53):
When I think about an agent,you've got a completely
autonomous service right andthen you can have an assistant
which helps augment, like I wastalking about a few minutes ago.
What do you think about thedifferences there?
Speaker 2 (16:13):
at control versus productivity.
So, like, assistance are saferfor, like, um, compliance, um,
you know, I think, capital, youknow, uh, dominion, like, you
know those big fortune 500companies, you know I think
you're going to probably want touse assistance with that, so
there's some augmentedintelligence in there.
But agents unlock efficiency byreducing human oversight.
(16:34):
So you know, I think, again,with the companies, I just with
(16:54):
assistance, just to make surethey can turn on and off the
autonomy.
And then, you know, as thistechnology gets a little bit
easier to refine, you know, um,and the guardrails are situated.
Then you know, I think, uh,it's a little bit easier.
But like, yeah, assistance islike, um, you steer, it replies,
and then agents are hey, I wantyou to book this trip to maui
for me and my wife and uh, makesure that we're on the beach.
(17:18):
You know, I think salesforce isreally have.
They have a really big pitchright now to get their agent
force out there.
So, like, from, from apractical, currently in the
market kind of thing, if you'rean enterprise looking to kind of
, you know, use, use, agents inparticular, I know Salesforce is
a, is a company that's reallypushing a lot of time and
attention.
And then you know a lot ofthese other companies are going
(17:39):
to get there as well.
But you know, I think I see alot of when I'm watching you
know market stuff.
I'm seeing a lot of agent talkcome from Salesforce in
particular.
Speaker 1 (17:47):
Yeah, I remember having this discussion with Will
Melton, who's a board member ofAI Ready RVA.
In one of the earlier episodeshe was telling me about
Salesforce, and this was severalmonths ago, but they really had
a huge push, like you said, andthey're super disruptive in the
market, so that was a reallygood example.
(18:08):
Yes, so now that we've gottenour audience to think about you
know what the agents are, howthey're being used, supporting
knowledge, workers in theirroles and some of the
differences between what anagent is and what an assistant
(18:28):
can be.
That leverages agentic AI.
What do you think about some ofthe risks and challenges, maybe
even some ethicalconsiderations when thinking
through how to use these agents,or maybe even where to start?
Speaker 2 (18:44):
I mean when you think of ethics, you know you already
talked about them earlier, soI'm going to shout them out.
Hopefully they can sponsor ourepisode.
I always go with Anthropic.
You know their whole entireethos is to like, not train on
your information, you know, andthey want to make sure that you
know everything is designed andconstructed to reduce, like,
(19:05):
hallucinations and misalignment.
And I always like, when itcomes to ethics, when I hear
that word, I think aboutanthropic 100%.
But you know safety frameworksare there.
You know OpenAI, for example.
You know they have.
You know, tools likemoderations for APIs to filter
harmful outputs.
(19:27):
But you know you still got tomake sure that you know you're
watching out for hallucinations.
I've seen stuff where you know,like I said earlier, I said, hey
, book a flight for me and mywife to Maui.
But when they book it, theybook a non-existent flight and
you know you fly from, like youknow, richmond, virginia to
Henrico and it's like, oh, thatdoesn't even make sense type of
deal, right.
So you know, stay on top of it,just from a risk mitigation
(19:49):
standpoint.
But if ever there was, you knowa company that I can always,
you know, give their flowers inthe moment.
When it comes to ethics, Ialways say Anthropic Very good.
Speaker 1 (19:59):
Very good.
So they're like the Volvo ofthe car industry.
Speaker 2 (20:04):
That's a good one True story about me.
My first car was a Volvo 960 athigh school.
You know my mama still, shestill gots a Volvo to this day.
So yeah, shout out Volvo, yep.
Speaker 1 (20:17):
Those marketing agencies train us well, don't
they?
Yeah, all right, man, let'stalk about agent-to-agent
interaction.
We'll call it multi-agentsystems, when agents can talk to
each other.
Tell me, what do you see inthere?
Speaker 2 (20:36):
So I had to do some research on this, just so I can
have something to speak to,because that's the first time I
heard about it, so I'm going toread from it and then we can
kind of speak to it.
But apparently Stanford'sSmallville showed 25 independent
agents forming social behaviorslike birthdays, restaurants,
meetups, without any script froma business angle, fleet
(20:59):
scheduling, supply chain biddingand even decentralized finance
bots that negotiate spreads inreal time, which is I'm hearing
that if I'm, if I'm gettingpitched a job and I have to
negotiate against a bot to getmore money, that's, that's going
to suck, man.
I need that human interaction.
Even, um, outside of that, I'llsay that, uh, decentralized
(21:23):
finance, so DeFi bots, you know,trading autonomously, um, like
multi-agent, and that's actually, you know, you know I'm a I'm a
big finance nerd, you know,once I can get that going like
that, that's a really good usecase to follow the market,
especially with this month.
Whenever this comes out, youknow, I think April is going to
probably be one of the morevolatile months that we've seen
(21:43):
in the market in decades.
Easy, so like having an AIagent, you know, to work
simultaneously with one anotherto kind of, like you know,
follow trends, definitelydefinitely a good way to
leverage multi-agent systemsabout that because I think it's
(22:14):
the pinnacle of agent-to-agentinteractions.
Speaker 1 (22:15):
Right?
You said that there were 25agents that were put into this
virtual environment calledSmallville and, from my research
, each of those were given apersonality and a little bit of
direction of how to start theirrole.
And then, when placed into thisenvironment, right, imagine a
(22:35):
little township, if you will,where there are multiple
buildings and 25 little avatarsrunning around, right?
Like, just think about whatthey could be doing if, given
the autonomy to go and, you know, just live out their day
(22:58):
however they wish, right?
So they'll wake up and observetheir environment, right, like
we talked about, and thenthey'll figure out, you know, do
they have a job, so should theygo to work?
That's their reasoning, right?
And then they leverage thememory of their previous day's
(23:21):
context in making thosedecisions as well.
So was yesterday Friday?
Then they don't have to go towork on a Saturday, right?
Maybe not anyway, do?
Do they want to go call uptheir friend and get some coffee
and meet at the coffee shop?
That sort of thing is reallyinteresting about observing this
(23:42):
environment.
Let's just take a look at itreal quick.
Go to Smallvillecom and take alook at that demo.
Here you go, screen share.
Go to Smallvillecom and take alook at that demo.
Speaker 2 (23:54):
Here you go screen share.
Actually, that's a good idea,you know.
I mean I'm trying to find thisso I can speak to it in the
moment, but I promise there wasa Black Mirror episode just
about this man.
Speaker 1 (24:09):
Oh, my God, yeah, so I watched it last night.
Speaker 2 (24:13):
Oh, yeah, okay, so you're talking about the one
with the guy with the sound atthe end, or whatever.
Speaker 1 (24:17):
Uh-huh.
Speaker 2 (24:18):
Yeah, I'm trying to find the name of it.
Yes, yes.
Speaker 1 (24:29):
He's talking with little agents that are running
around living their life in anepisode like this, and then
they're speaking to him.
They tell him how to build acomputer system that allows them
to become more powerful, and Idon't want to spoil Spoiler
alert right.
Yeah, I know, if you like BlackMirror, don't listen to this
(24:51):
part, but the little agents thatare running around are given
autonomy to escape into theworld through some anomaly at
the end of the show, and thenthey can project their sound
across all the humanity andstart controlling them.
It looks like that's where theepisode ends, so you have to
(25:12):
kind of use your imagination,but they look a lot like this.
You know, you see all theselittle guys running around.
They have purpose, right?
They've figured out how to givethese agents purpose, so you
click on one of them and you canfind out where they are.
It looks like the demo hasended here.
(25:36):
For me personally, because I'vebeen running it for so long, I
see out here that they'reselling NFTs too, which is
always fun.
And shout out to my Web3 people.
Speaker 2 (25:48):
I'm a Web3 guy man, but you can never not have an
NFT hustle right.
This is amazing, though, andthen it's just crazy, because
whenever I watch Black Mirror,it's crazy how cinema sometimes
is reality.
I remember a couple seasons agowhere they had like a social
score episode oh my god that onewas crazy.
(26:11):
But if you look at somecommunist countries in the world
, they have that going.
Yeah, exactly, and I won't saythe names because I don't want
to get demonetized over there.
I hope y'all listening fromover on the other side of the
world, but that's a real thing.
So it's like, yeah, like that,that episode, this is literally
the framework of it, and youknow what it reminded me of?
I don't know if you're a oldschool PC gamer, but like doom
(26:31):
and like Duke Nukem, you know it.
It.
Speaker 1 (26:35):
I love it that I was raised on Doom.
Speaker 2 (26:39):
Oh man, Like the user interface looked just like it.
So it's like it's funny howstuff comes full circle.
Speaker 1 (26:43):
Yeah, yeah, that's right.
Okay, where were we Having funtalking about tech stuff, all
right.
So Smallville, black Mirror,some really exciting but can be
scary stuff let's think aboutthe future, right, like, let's
be practical about this.
Let's think about the future,right, let's be practical about
(27:24):
this, not have to spend as muchtime doing X, y or Z.
So tell me what are yourthoughts on personal AI agents?
Speaker 2 (27:43):
statistics they're suggesting and they're
predicting a quarter of digitaltasks are going to be done by
personal agents.
I guess this time next year Ithink it's interesting If I
could have my own personal chiefof staff that just takes over
scheduling emails for me and,again, finance and meal prepping
and just making sure that Igive my mom an excellent
Mother's Day gift, because I'mterrible at that stuff.
(28:06):
I think that it's awesome.
But there's two sides to that.
Right, I'm a white hat kind ofguy.
Black hats though I mean itjust makes it way easier to scan
.
You know the way easier to use.
You know technology for war.
Know it's a lot of a lot ofstuff going on, whether it's in
um in gaza.
(28:27):
You know everything that'sgoing on over there.
And then you know I saw russiaum just retook some um land that
ukraine um had held for a longtime, right.
So, like you know, when it,when it comes to these agents,
it's like a, it's a double-edgedsword because you know, I think
you and I we're pretty when itcomes to this stuff.
Like you know we're whiteheight kind of guys, but like
protecting from that um, it'sgoing to be something that I
(28:48):
think whoever could come up witha good use case to kind of make
sure that we don't blow theworld up with this type of thing
.
Man, um, I think it's gonna begood, but it's like you know, I
always think um two sides of thecoin.
You know, personally it's goingto help me be a lot more
efficient, but it's just likewhere I'm at in my life 20 years
ago, victor.
(29:08):
I'm thinking somethingdifferent.
I'm like, yeah, how can theyhelp me cut the line to get into
the next big party, or stufflike that?
I think it's going to adaptwith all the users and this is
going to adapt with scale.
Palantir again, company, youknow, check them out.
Um.
Wouldn't surprise me if theystart um using um, you know, ai
(29:31):
agents for defense, or ifthey're already doing it now.
So I mean, the future is, uh,it's wide, open, man, wild, wild
west out here when it comes tothis.
I think.
Speaker 1 (29:40):
Yeah, I think companies like Rabbit are close
to making it more availablescalable Rabbit but for those
who don't know what that is,there's a little device, much
like a mobile smartphone, thatcan execute agentically execute
tasks for you.
(30:01):
So giving more freedom topeople through personal
assistance, in my mind, iscreating packaged agents that
are specialized in doing thingsright, like I'm sure it's going
to be really challenging tocreate agents that can do
anything for people right, likethat's just a lot of
(30:23):
coordination and extremelyexpensive to create generalized
agents like that, but I dobelieve that people are gonna be
able to pick and choose agentsin the future to be able to.
You know, maybe it's like asubscription.
You know, I want an agent.
(30:44):
I want to pay monthly fees tothis agent that will write all
my emails for me, or I want thistravel agent, agentic AI, to be
able to book all of my work andmy personal vacations for me.
Those are just some things, butI do think that they need to be
(31:06):
specialized, at least right now,because we don't have
artificial general intelligenceto be able to think through and
piece together on its owneverything that it needs to make
choices in a world as complexas ours.
But the next iteration issomehow packaging up the next
(31:29):
Alexa right and making it apersonalized assistant to
whatever suits your fancy.
Whatever you think is going tosupport your needs the most, and
I think that's where the scaleis going to come from.
I think companies are going tosupport your needs the most and
and I think that's where thescale is going to come from I
think companies are going to befinding ways to package those up
and ship them off and make lotsand lots of money from them um
(31:51):
cognition labs is actually doingthat.
Speaker 2 (31:53):
They're a sas platform.
I guess it's like an agent as aservice as opposed to sas, but
like um, that's that's a gooduse case.
And then um always follow themoney.
You know, if you're you'relooking at the sequoias and the
vc um trends right now, theangel investors you know, agent
startups are attracting fundingum at a pretty pretty high rate
(32:14):
right now.
Um, so you know, check thoseout.
And then another thing too likesolopreneurs are going to be
able to really take advantage ofthis.
Just like, imagine like a, aone-person startup in richmond
that can have their own digitalworkforce.
And like fix stuff that needsissues like the water system.
Like for for those listeningfrom the other side of the world
(32:34):
, like at the start of 2025,richmond and ryko and the
surrounding areas had no waterright, what if there was an
individual that can get ahead ofstuff like that to anticipate,
you know, water droughts or whatended up happening with that
situation?
I think there was like amachine function messed up and
then it flooded somethingsomewhere and then it just
(32:55):
messed up everything right.
But like, I think, the AIagents and individuals with the
passion and the energy to kindof really, you know, take that
thing and run with it are goingto be able to do it by
themselves one person and thencome up with some great
solutions to really have apositive effect on the community
.
Yeah, definitely.
And just because we're here,will Milton shout out RVA Water.
(33:19):
You know he did a lot of goodstuff when we were short on
water in the richmond area.
So if you're ever in richmondand you see a cool little uh
aluminum bottle, that's one ofour uh founders.
Uh, out here they already.
Uh will milton um giving uhgiving water.
And great guy, great, greatcompany.
The water's way better than thestuff that you would get out of
aquafina or a designing.
(33:39):
So so you know.
A little.
Public service announcement.
Speaker 1 (33:43):
All right Go Will.
All right man.
Speaker 2 (33:51):
What do you think the next few years looks like for
AI agents, if the trajectory isgoing to go like the same hype
cycle with, like just AI ingeneral, I say what OpenAI
released, release chat, gpt inNovember of 22.
And then by the end of 23, forthe early adopters, I was in
conferences listening to AI andthen at the end of last year I
(34:14):
mean everybody is aware of it solike in two years, if it took
us to get there, I think AIagents probably get there in six
.
I think you know assistancewith you know agents is going to
start to really affect a coupleof industries.
So if you're in theseindustries, try to figure out
how you can show your employerthat you're using AI to help you
be more efficient.
(34:35):
So, marketing, coding,technology, all that, and then
customer support, I think I knowsome call centers are already
starting to introduce callcenter agents and things of that
nature.
So, like you know, make surethat you know.
You just kind of you know, geta little bit of knowledge
underneath your belt so you canbe ahead of the competition.
You know, because if they haveto cut staff, you might want to
(34:57):
cut the staff who has nounderstanding of what's going on
with AI agents versus theperson who's always talking
about hey, I saw this researcher.
I'm thinking this solution isbetter, right?
So I think, in about like sixmonths, what's the April?
Yeah, about the end of the yearit's going to be a lot more
plug and play, no code, easy toexecute and implement AI agents,
and I think it's just becausewhen you looked at where just
(35:19):
artificial intelligence came, ittook about like two years for
it to just be like.
I mean, like you can't goanywhere without anybody talking
about ai agents now, and Ithink that they've been talking
about ai agents for the earlyadopters at the end of last year
, for sure, and it's starting tokind of pop up a little bit
more.
So, six months, for sure, andthen long term, like five years
from now, the aging ecosystem isgoing to be like um, that's a
(35:40):
different conversation.
You know that we should have,but like, I think it's is going
to be like that's a differentconversation you know that we
should have, but like I thinkit's just going to be a
different use case for people towork.
I look forward to seeing how Ican compete in that market.
You know like I'm trying tostay as knowledgeable as
possible.
So I'm going to the event onTuesday and you know anytime
that if you're interested andwant to get a cup of coffee, you
know, check me out onSustainable creative.
You know, I love just havingthese conversations.
(36:01):
I love hopping on with jasonyou know it's early, he got, he
got to get the garage knockedout, I got to cut the grass.
But it's like whenever you taketime out to like, you know,
just be humble and haveconversations like this, you
just get better.
So I say, um, you know, sixmonths, it's going to be a
little bit quicker mass adoption, but in, like you know, five
years, um, it's, it's going tobe hard to predict.
I mean, differentadministration, different world,
(36:23):
it's going to be interesting.
Speaker 1 (36:25):
Yeah, maybe different administration.
I like it.
Look, I would love to have thatconversation.
We're not having thatconversation.
2028 is very unpredictable.
Speaker 2 (36:37):
Yeah, but.
Speaker 1 (36:38):
I do think that you're right on.
With the vast change that'scoming, it's hard to tell what
agent to agents are going to beable to do and what big tech is
going to do with these thingsand how they're going to mass
produce them and roll them outinto society.
Right, it's like you'reabsolutely right.
It's like do what you can todayto learn and grow with these
(37:01):
technologies, because there's nogetting around it.
They are in our world andthey're going to become more
prevalent, more available.
Get to know them and don't beafraid of them.
Right, don't be afraid of thetechnology and trying new things
with them and seeing if you canexplore different means of
leveraging them so that they'revery approachable in your work
(37:24):
and life.
And when your company decidesto start leveraging this stuff
internally, you're ready for it.
But we're here to help AI ReadyRVA.
That is our mission.
We want to support thiscommunity.
So let us know how we can Reachout to us in LinkedIn.
Reach out to our friends atSustainable Growth Creative.
(37:45):
Reach out to them and let themknow how they can help you.
That's their business model.
They want to bring people up tospeed.
We also have a really goodplatform that we're developing
at AIReadyRVAcom.
It's going to help educatefolks.
So become a member today and wewill be contributing in great
(38:07):
force to the growth and scalingeducational opportunities for
our cohorts and everyone, allour members.
So join us online and get toknow AI today, because the
future is not coming, it isalready here.
Thanks again, vic.
Speaker 2 (38:25):
Well, actually, can I close this For people who stay
to the end of this podcast, Igot something bonus.
Can I do a bonus thing realquick?
Yeah, do it.
Have you heard of Manus AI yet?
Speaker 1 (38:35):
Yes, yeah, it's a Chinese startup, right.
Speaker 2 (38:39):
Yes, it is Just like DeepSeek, and that's another
thing.
I think that the um you know,the chinese are um, really like
what they did with leveragingolder um.
They weren't even blackwellgpus from nvidia, what they
actually did, and it's super inthe weeds um, and I'm open if
somebody in the comments cancorrect me if I'm wrong but like
(39:01):
all those GPUs from NVIDIA,they use CUDA as like their
programming language.
Apparently, deepseek figuredout how to get around that and
that's how they were able to getthose chips to function so
effectively.
But saying that, to say thisthat Manus AI is just as
disruptive as DeepSeek.
If ChatGPT is an assistant,manus is a project manager that
(39:26):
actually does the project workfor you.
Check them out.
They're a state of the art GIAagent benchmark and all that is
is just like measuring success,like they benchmark all these
tools and like Manus AIapparently scored pretty high on
(39:48):
that.
Check them out.
And then, from BC's 75 million,what a $500 million evaluation.
So it's not like they're a toolor a company that is not
turning heads.
That's a pretty high evaluationand that's.
You know.
You follow the dollars.
You know I was able to use itto come up with I'm doing a
score three-part AI seriescoming up.
(40:12):
I should probably pump that, butI ain't got the dates yet.
But it's a great tool.
It's using a whole bunch ofdifferent AI agents at one time
and it helped me come up with agood curriculum and it took me
maybe like 10, 15 minutes to getin there.
And then I just you know, tookthat information, put it in
gamma and I got I got a hell ofa presentation right.
So if you're looking to try tosee like what's the example of
(40:34):
what an AI agent can do rightnow, in the moment you can
download Manus AI and they got abeta version.
But just be particular, becauseit's just like Deep Seek.
That stuff is going to be on aChinese server.
And if you don't want yourstuff on a Chinese server,
there's a multitude of reasonswhy you wouldn't.
I'm not going to get into them.
Just make sure that you're kindof particular about what you
put on there.
I'm asking it to help me witheducational information and the
(41:00):
stuff they give back is legit.
I'm telling you there's goingto be a Chinese Coca-Cola coming
out if you do that.
So don't do that right, becauseI mean like this is the end of
the podcast and, like I alwayslike to reward people who listen
to the end, give them some youknow little Easter eggs to kind
of go back and, you know,hopefully talk about the stuff
in the comments.
Speaker 1 (41:15):
Awesome Thanks.
So you have a wonderful rest ofyour weekend and I look forward
to catching up with you againsoon.
Speaker 2 (41:22):
Yes, sir.
Speaker 1 (41:26):
And thanks to our listeners for tuning in today.
If you or your company wouldlike to be featured in the
Inspire AI Richmond episode,please drop us a message.
Don't forget to like, share orfollow our content and stay up
to date on the latest events forAI Ready RBA.
Thank you again and see younext time.
Welcome RVA to Inspire AI, where we spotlight
companies and individuals in theregion who are pioneering the
development and use ofartificial intelligence.
I'm Jason McGinty from AI ReadyRVA.
At AI Ready RVA, our mission isto cultivate AI literacy in the
greater Richmond region throughawareness, community engagement
(00:24):
, education and advocacy.
Today's episode is madepossible by Modern Ancients
driving innovation with purpose.
Modern Ancients uses AI andstrategic insight to help
businesses create lasting,positive change with their
unique journey consultingpractice.
(00:44):
Find out more about how yourbusiness can grow at
modernagentscom, and thanks toour listeners for tuning in
today.
If you or your company wouldlike to be featured in the
Inspire AI Richmond episode,please drop us a message.
Don't forget to like, share orfollow our content and stay up
(01:07):
to date on the latest events forAI Ready RVA.
All right, welcome back toInspire AI.
Today we have an openconversation with Vic Rogers
from Sustainable Growth Creative.
He's a regular on the show.
We've had him talk about hiscompany, his role in AI Ready
(01:28):
RVA as a board of directors andhis enthusiasm for AI agents,
which is the future ofartificial intelligence,
autonomous nature, and here weare to discuss some of our
recent findings and to create aspace where this is just an
(01:48):
approachable share out of ourlatest thinking, so we want to
talk a little bit about what'sthe latest in the technology,
not so much in a deep dive oftechnical components, but how
does this stuff work, where isit used, and we'll get into some
of that.
So welcome back and it's greatto have you on Vic.
Speaker 2 (02:10):
Oh yeah, man, Looking forward to it.
I think this is goodPre-homework before the event on
Tuesday, with the AI agentsession going on, so I'm looking
forward to getting a little bitstronger in this subject matter
.
Oh, yeah, tell us a little bitabout that, yeah, so, depending
on when this comes out, I'm notsure which cohort that's coming
(02:30):
under, but there's going to be,on April 29th, an AI agents what
they are and what they aren'tsession going to be ran by a
corporation called Rise andScaleai and it's going to be a
Dominion payroll from 530 to 7.
It's one of those things where,if you're interested in just
(02:51):
understanding what AI is, we gota heavyweight coming in, a
young lady by the name ofJessica Clark who's a senior
solutions architect at NVIDIA.
She's going to be doing a lotof education.
I think it just goes well.
This is just a pre-work beforethat, or, if you listen to it
afterward, you know same deal.
You know lock in with what theygot going on over there support
, support the sponsor, rise inscale, and you know if you can,
(03:23):
if you ever get a chance to talkto somebody from NVIDIA about
this stuff.
You know I think right nowthey're at a pace center when it
comes to like being ahead in amarket with it's, with this
information.
So I've been doing a littleresearch just to make sure when
I go in and this doesn't go allover my head good, yeah, well,
uh, dominion payroll.
Speaker 1 (03:32):
Tuesday evening, nvidia's architect, jessica
clark, be giving us a greatpresentation.
I don't have any guaranteesthat this episode is going to
come out before then, but getyour tickets.
Yes, sir.
All right, follow us onLinkedIn AI Ready RVA for the
latest news as well as events.
(03:54):
So let's get this conversationstarted, vic.
Let's talk about what AI agentsreally are.
What do you think they are, vic?
Speaker 2 (04:06):
I think it's software that perceives reasons and acts
towards a goal.
It's a little bit more thanwhat you see with the actual LLM
.
You know, when you think ofLLMs, you know they're taught in
their power from an externaldatabase.
You know a lot of these agents.
They're going to be rules-basedbots and they're going to have
(04:28):
scripts and those agents aregoing to be able to adapt in
real time, which is amazing,right?
So I personally am lookingforward to like this year being
the year of agentic AI.
You know that term just soundsfunny, but that's what the
Zeitgeist is.
You know, using agentic AI.
You know that term just soundsfunny, but that's what the
Zeitgeist is using agentic AI.
(04:48):
And you know, when I did myresearch, that's how I came up
with it.
What about yourself?
What are your thoughts?
Speaker 1 (04:54):
Yeah, I think agents are advanced autonomous systems,
right, like.
You have an LLM that is kind ofthe brain of the agent, and
then the agent is the programthat leverages the LLM and it'll
leverage other tools Maybe wecan talk about some of those
(05:15):
tools as well to autonomously,like you said, reason about the
environment.
So it observes and then itreasons and then it takes action
, right, like.
I think that's what the at theat the end of the day, the crux
of the of what an AI agent is,it's able to observe, reason and
then act upon that, and thethinking part of it right is is
(05:39):
really just about, like, givingit context.
So the LLM itself, theabundance of information in the
internet, right, can be used inthe agent's observations and
reasoning.
Right, it can think about whatkind of information it's
available to leverage and how itmight take what it's observed
(06:03):
and think a little bit morestrategically or tactically
about what its task is.
So there you go.
I think those are the types ofthings that jump out at me when
I think about it.
I think there's a lot ofapplications to AI agents.
You know you got your robotics,you've got your customer
(06:24):
service support centers and whatelse?
Where else might you see agentsthese days, vic?
Speaker 2 (06:31):
Customer care, 100%.
You know I think that callcenter work is going to be
replaced pretty soon with a lotof agents.
And I know, don't quote me onthis one I can't remember the
actual fast food company, but itmight've been Wendy's actually,
or I might be confusing Wendy'ssuper side note, like with
(06:54):
Palantir.
I know Palantir and Wendy's areworking one another and that's
kind of weird.
But just having, I think at onepoint that there was a goal to
try to use ai to take drop totake menu orders, you know, and
it didn't work out.
But like I think they'rethey're refining that a little
bit better.
But then also with likemarketing, like right now, you
(07:15):
know, um, you can have a oneperson news crew where you know
you can use the ai agents tocreate blog posts and then just
make sure the seo is where itneeds to be and then poof, you
know you're out there with somegood content.
I think the one that's probablythe scariest, especially for,
like time spent in the space,software development.
Like you know, if agents canstart coding, like you know,
(07:37):
it's kind of, like you know,tough for, like you know,
developers, coders, architects,who have been in the space for a
long time to have thistechnology come through because,
like, what's going to happen islike I don't know if I say one
in four or you know it's a nicejump, but like you won't
necessarily need to have thestaff to hire as many of these
architects and coders anymorebecause you could just have an
ai agent paired with a human todo the same amount of work 10
(08:00):
times faster.
Right um, weeks turn, turn intodays when you're looking into
that Finance too.
You know just a lot ofdifferent use cases and you know
, I would say you know picturinglike nonprofits, getting into
leveraging.
You know these agents withouthaving to hire, like data
science teams, you know.
So there's a lot of good and alot of displacement, but if used
(08:22):
properly in the rightindustries, the right use cases,
I could see a lot of positivehappening.
Speaker 1 (08:26):
Yeah, yeah, I think we'll have an opportunity to
talk about AI autonomous humanreplacement versus augmentation.
I think that we're on a path toaugment right now.
Full replacement isn't so muchin the short term for us right
now, because I think thatthere's still some concerns
(08:49):
about the reliability of itsoutputs Right, and nobody wants
to be put on the front page for,you know, last week releasing
an autonomous agent and thenthis week, you know, blowing up
their industry, right?
So I think there's there'sstill going going to be some
work to do before fullreplacement.
But, yes, augmentation, as wesay, is definitely very
(09:13):
approachable for companies andthey are considering that at
full speed.
So we've gotten to the pointwhere we know what an AI agent
is, you know, logically speaking, and we know how it's being
used in companies, industries.
What about the building blocksof AI agents?
(09:35):
Let's click on that one alittle bit.
What do you think are importantcomponents to creating an agent
and making it successfullyoperate?
Speaker 2 (09:48):
I guess a couple of things.
I think a good analogy I sawone time was that you know AI
agents are like Lego sets, likeLLMs are leveraged for thinking
and then tools for acting.
You know memory for learningand then planning for strategy.
So like that's a good kind oflike way to you know, think
about it.
Or even when you think aboutthe building blocks, you know
(10:09):
you got a brain, you got atoolbox, you got a diary, you
got a whiteboard and then youwire them all together and bow,
you got this beautiful AI toolright.
So just broken down and like alayman term, I think that's
that's the easiest way to kindof like fill it out when it
comes to like the buildingblocks of.
Speaker 1 (10:26):
Okay, there are some, definitely some tools out there
.
Let's let's take, for instance,an agent that can scrape any
website and help formulate anopinion of you know what a
company does, how your businesscan support it, that sort of
thing.
When I think about that usecase, this agent could be using
(10:48):
a, you know, browser type searchtool, a scraper, something that
will pull information fromspecific website or web pages
and then some sort ofsummarization capabilities with
the LLM, then use some sort ofRAG-like so retrieval,
augmentation, generation,RAG-like thought process to
(11:12):
review, based on your companyopportunities or service-level
opportunities for your companyto help with and possibly make a
pitch for.
So all of that is the types oftool sets that agents can use in
(11:51):
just that one circumstance.
I think that I want to talk alittle bit more about how it
does that.
So humans use memory to observeand reflect on things.
Agents do the same thing, right.
Agents do the same thing, right.
So in that case, how do youthink that the memory would
(12:12):
surface and be of support inthat structure?
What do you think about that,Vic?
Speaker 2 (12:19):
Vector DBs such as Pinecone and Prestigious can
keep past calls so the agentscan actually remember your
specific use cases, like if it'slike organizational policies
and things of that nature.
I think from a memorystandpoint, you know that's
(12:41):
ideal Persistent storage, justenabling agents to recall past
interactions and improve overtime and like build over time.
I think both of those are umexamples of how like memory is
going to be leveraged.
When it comes to using these umai agents and actually I was
gonna say in the future, butlike now, currently like there's
some tools out there that we'llprobably talk to eventually I
(13:02):
haven't seen a lot of like nocode easy to use ones yet I've
been playing trying to figureout some for a client of mine.
But um you, it comes to memory.
That's what I'm seeing some ofthe best speak to.
Speaker 1 (13:16):
The Pinecone one I've been using recently in building
a chatbot, just testing it outand seeing how scraping a
website and storing it into thevector database, which is
PinecCone right, how that works,being able to access that
(13:36):
memory, if you will, from thechatbot's interface, asking the
bot to recall what it learned byjust submitting some basic
questions about the website thatit ingested, see what it knows,
and so far it does reallyreally well.
So a little bit of explorationthere goes a long way, for sure.
Speaker 2 (14:01):
How long have you been doing that too?
That sounds kind of cool.
So you're already in there,kind of coding an AI agent.
Speaker 1 (14:08):
Well, actually I used Claude Claude 3.7, Sonnet, I
think it's what it's called andjust the free version if you, if
you know how to instruct it tocreate something like that's.
What I've been doing lately isjust testing out its
capabilities for producing anentire application like that.
(14:29):
I'm happy to give you a demoafter the podcast, but it's
pretty amazing to be able to seewhat it can produce.
And the problem with it is thatif you have it build something
that's extravagant and youactually get it to work right,
but you don't really know howit's working, then you can't
(14:53):
really update it right, youcan't maintain it.
You ask it to do something elseagain, to try to like adapt it
or create a you know, a newfeature or something like that,
and it'll go back through andrewrite the entire code, which
changes things right, and youhave no idea what it's changing.
So that becomes one of theblockers for it these days.
Speaker 2 (15:18):
But I'm happy to talk with you a little bit more
about that after the episodeDefinitely one of the benefits
of being a co-host getting coolstuff like that man.
So look forward to thatconversation.
Speaker 1 (15:30):
Yeah, definitely getting getting cool stuff like
that man.
So look forward to thatconversation.
Yeah, definitely, um, it was.
It was kyle.
Uh, johnson from our platforms,who you know shared with me his
insights into claude and how hewas using it and I was like I
need to go try that.
And so there you go.
Um, tell me what.
What are your thoughts on theagent versus assistant debate?
(15:53):
When I think about an agent,you've got a completely
autonomous service right andthen you can have an assistant
which helps augment, like I wastalking about a few minutes ago.
What do you think about thedifferences there?
Speaker 2 (16:13):
at control versus productivity.
So, like, assistance are saferfor, like, um, compliance, um,
you know, I think, capital, youknow, uh, dominion, like, you
know those big fortune 500companies, you know I think
you're going to probably want touse assistance with that, so
there's some augmentedintelligence in there.
But agents unlock efficiency byreducing human oversight.
(16:34):
So you know, I think, again,with the companies, I just with
(16:54):
assistance, just to make surethey can turn on and off the
autonomy.
And then, you know, as thistechnology gets a little bit
easier to refine, you know, um,and the guardrails are situated.
Then you know, I think, uh,it's a little bit easier.
But like, yeah, assistance islike, um, you steer, it replies,
and then agents are hey, I wantyou to book this trip to maui
for me and my wife and uh, makesure that we're on the beach.
(17:18):
You know, I think salesforce isreally have.
They have a really big pitchright now to get their agent
force out there.
So, like, from, from apractical, currently in the
market kind of thing, if you'rean enterprise looking to kind of
, you know, use, use, agents inparticular, I know Salesforce is
a, is a company that's reallypushing a lot of time and
attention.
And then you know a lot ofthese other companies are going
(17:39):
to get there as well.
But you know, I think I see alot of when I'm watching you
know market stuff.
I'm seeing a lot of agent talkcome from Salesforce in
particular.
Speaker 1 (17:47):
Yeah, I remember having this discussion with Will
Melton, who's a board member ofAI Ready RVA.
In one of the earlier episodeshe was telling me about
Salesforce, and this was severalmonths ago, but they really had
a huge push, like you said, andthey're super disruptive in the
market, so that was a reallygood example.
(18:08):
Yes, so now that we've gottenour audience to think about you
know what the agents are, howthey're being used, supporting
knowledge, workers in theirroles and some of the
differences between what anagent is and what an assistant
(18:28):
can be.
That leverages agentic AI.
What do you think about some ofthe risks and challenges, maybe
even some ethicalconsiderations when thinking
through how to use these agents,or maybe even where to start?
Speaker 2 (18:44):
I mean when you think of ethics, you know you already
talked about them earlier, soI'm going to shout them out.
Hopefully they can sponsor ourepisode.
I always go with Anthropic.
You know their whole entireethos is to like, not train on
your information, you know, andthey want to make sure that you
know everything is designed andconstructed to reduce, like,
(19:05):
hallucinations and misalignment.
And I always like, when itcomes to ethics, when I hear
that word, I think aboutanthropic 100%.
But you know safety frameworksare there.
You know OpenAI, for example.
You know they have.
You know, tools likemoderations for APIs to filter
harmful outputs.
(19:27):
But you know you still got tomake sure that you know you're
watching out for hallucinations.
I've seen stuff where you know,like I said earlier, I said, hey
, book a flight for me and mywife to Maui.
But when they book it, theybook a non-existent flight and
you know you fly from, like youknow, richmond, virginia to
Henrico and it's like, oh, thatdoesn't even make sense type of
deal, right.
So you know, stay on top of it,just from a risk mitigation
(19:49):
standpoint.
But if ever there was, you knowa company that I can always,
you know, give their flowers inthe moment.
When it comes to ethics, Ialways say Anthropic Very good.
Speaker 1 (19:59):
Very good.
So they're like the Volvo ofthe car industry.
Speaker 2 (20:04):
That's a good one True story about me.
My first car was a Volvo 960 athigh school.
You know my mama still, shestill gots a Volvo to this day.
So yeah, shout out Volvo, yep.
Speaker 1 (20:17):
Those marketing agencies train us well, don't
they?
Yeah, all right, man, let'stalk about agent-to-agent
interaction.
We'll call it multi-agentsystems, when agents can talk to
each other.
Tell me, what do you see inthere?
Speaker 2 (20:36):
So I had to do some research on this, just so I can
have something to speak to,because that's the first time I
heard about it, so I'm going toread from it and then we can
kind of speak to it.
But apparently Stanford'sSmallville showed 25 independent
agents forming social behaviorslike birthdays, restaurants,
meetups, without any script froma business angle, fleet
(20:59):
scheduling, supply chain biddingand even decentralized finance
bots that negotiate spreads inreal time, which is I'm hearing
that if I'm, if I'm gettingpitched a job and I have to
negotiate against a bot to getmore money, that's, that's going
to suck, man.
I need that human interaction.
Even, um, outside of that, I'llsay that, uh, decentralized
(21:23):
finance, so DeFi bots, you know,trading autonomously, um, like
multi-agent, and that's actually, you know, you know I'm a I'm a
big finance nerd, you know,once I can get that going like
that, that's a really good usecase to follow the market,
especially with this month.
Whenever this comes out, youknow, I think April is going to
probably be one of the morevolatile months that we've seen
(21:43):
in the market in decades.
Easy, so like having an AIagent, you know, to work
simultaneously with one anotherto kind of, like you know,
follow trends, definitelydefinitely a good way to
leverage multi-agent systemsabout that because I think it's
(22:14):
the pinnacle of agent-to-agentinteractions.
Speaker 1 (22:15):
Right?
You said that there were 25agents that were put into this
virtual environment calledSmallville and, from my research
, each of those were given apersonality and a little bit of
direction of how to start theirrole.
And then, when placed into thisenvironment, right, imagine a
(22:35):
little township, if you will,where there are multiple
buildings and 25 little avatarsrunning around, right?
Like, just think about whatthey could be doing if, given
the autonomy to go and, you know, just live out their day
(22:58):
however they wish, right?
So they'll wake up and observetheir environment, right, like
we talked about, and thenthey'll figure out, you know, do
they have a job, so should theygo to work?
That's their reasoning, right?
And then they leverage thememory of their previous day's
(23:21):
context in making thosedecisions as well.
So was yesterday Friday?
Then they don't have to go towork on a Saturday, right?
Maybe not anyway, do?
Do they want to go call uptheir friend and get some coffee
and meet at the coffee shop?
That sort of thing is reallyinteresting about observing this
(23:42):
environment.
Let's just take a look at itreal quick.
Go to Smallvillecom and take alook at that demo.
Here you go, screen share.
Go to Smallvillecom and take alook at that demo.
Speaker 2 (23:54):
Here you go screen share.
Actually, that's a good idea,you know.
I mean I'm trying to find thisso I can speak to it in the
moment, but I promise there wasa Black Mirror episode just
about this man.
Speaker 1 (24:09):
Oh, my God, yeah, so I watched it last night.
Speaker 2 (24:13):
Oh, yeah, okay, so you're talking about the one
with the guy with the sound atthe end, or whatever.
Speaker 1 (24:17):
Uh-huh.
Speaker 2 (24:18):
Yeah, I'm trying to find the name of it.
Yes, yes.
Speaker 1 (24:29):
He's talking with little agents that are running
around living their life in anepisode like this, and then
they're speaking to him.
They tell him how to build acomputer system that allows them
to become more powerful, and Idon't want to spoil Spoiler
alert right.
Yeah, I know, if you like BlackMirror, don't listen to this
(24:51):
part, but the little agents thatare running around are given
autonomy to escape into theworld through some anomaly at
the end of the show, and thenthey can project their sound
across all the humanity andstart controlling them.
It looks like that's where theepisode ends, so you have to
(25:12):
kind of use your imagination,but they look a lot like this.
You know, you see all theselittle guys running around.
They have purpose, right?
They've figured out how to givethese agents purpose, so you
click on one of them and you canfind out where they are.
It looks like the demo hasended here.
(25:36):
For me personally, because I'vebeen running it for so long, I
see out here that they'reselling NFTs too, which is
always fun.
And shout out to my Web3 people.
Speaker 2 (25:48):
I'm a Web3 guy man, but you can never not have an
NFT hustle right.
This is amazing, though, andthen it's just crazy, because
whenever I watch Black Mirror,it's crazy how cinema sometimes
is reality.
I remember a couple seasons agowhere they had like a social
score episode oh my god that onewas crazy.
(26:11):
But if you look at somecommunist countries in the world
, they have that going.
Yeah, exactly, and I won't saythe names because I don't want
to get demonetized over there.
I hope y'all listening fromover on the other side of the
world, but that's a real thing.
So it's like, yeah, like that,that episode, this is literally
the framework of it, and youknow what it reminded me of?
I don't know if you're a oldschool PC gamer, but like doom
(26:31):
and like Duke Nukem, you know it.
It.
Speaker 1 (26:35):
I love it that I was raised on Doom.
Speaker 2 (26:39):
Oh man, Like the user interface looked just like it.
So it's like it's funny howstuff comes full circle.
Speaker 1 (26:43):
Yeah, yeah, that's right.
Okay, where were we Having funtalking about tech stuff, all
right.
So Smallville, black Mirror,some really exciting but can be
scary stuff let's think aboutthe future, right, like, let's
be practical about this.
Let's think about the future,right, let's be practical about
(27:24):
this, not have to spend as muchtime doing X, y or Z.
So tell me what are yourthoughts on personal AI agents?
Speaker 2 (27:43):
statistics they're suggesting and they're
predicting a quarter of digitaltasks are going to be done by
personal agents.
I guess this time next year Ithink it's interesting If I
could have my own personal chiefof staff that just takes over
scheduling emails for me and,again, finance and meal prepping
and just making sure that Igive my mom an excellent
Mother's Day gift, because I'mterrible at that stuff.
(28:06):
I think that it's awesome.
But there's two sides to that.
Right, I'm a white hat kind ofguy.
Black hats though I mean itjust makes it way easier to scan
.
You know the way easier to use.
You know technology for war.
Know it's a lot of a lot ofstuff going on, whether it's in
um in gaza.
(28:27):
You know everything that'sgoing on over there.
And then you know I saw russiaum just retook some um land that
ukraine um had held for a longtime, right.
So, like you know, when it,when it comes to these agents,
it's like a, it's a double-edgedsword because you know, I think
you and I we're pretty when itcomes to this stuff.
Like you know we're whiteheight kind of guys, but like
protecting from that um, it'sgoing to be something that I
(28:48):
think whoever could come up witha good use case to kind of make
sure that we don't blow theworld up with this type of thing
.
Man, um, I think it's gonna begood, but it's like you know, I
always think um two sides of thecoin.
You know, personally it's goingto help me be a lot more
efficient, but it's just likewhere I'm at in my life 20 years
ago, victor.
(29:08):
I'm thinking somethingdifferent.
I'm like, yeah, how can theyhelp me cut the line to get into
the next big party, or stufflike that?
I think it's going to adaptwith all the users and this is
going to adapt with scale.
Palantir again, company, youknow, check them out.
Um.
Wouldn't surprise me if theystart um using um, you know, ai
(29:31):
agents for defense, or ifthey're already doing it now.
So I mean, the future is, uh,it's wide, open, man, wild, wild
west out here when it comes tothis.
I think.
Speaker 1 (29:40):
Yeah, I think companies like Rabbit are close
to making it more availablescalable Rabbit but for those
who don't know what that is,there's a little device, much
like a mobile smartphone, thatcan execute agentically execute
tasks for you.
(30:01):
So giving more freedom topeople through personal
assistance, in my mind, iscreating packaged agents that
are specialized in doing thingsright, like I'm sure it's going
to be really challenging tocreate agents that can do
anything for people right, likethat's just a lot of
(30:23):
coordination and extremelyexpensive to create generalized
agents like that, but I dobelieve that people are gonna be
able to pick and choose agentsin the future to be able to.
You know, maybe it's like asubscription.
You know, I want an agent.
(30:44):
I want to pay monthly fees tothis agent that will write all
my emails for me, or I want thistravel agent, agentic AI, to be
able to book all of my work andmy personal vacations for me.
Those are just some things, butI do think that they need to be
(31:06):
specialized, at least right now,because we don't have
artificial general intelligenceto be able to think through and
piece together on its owneverything that it needs to make
choices in a world as complexas ours.
But the next iteration issomehow packaging up the next
(31:29):
Alexa right and making it apersonalized assistant to
whatever suits your fancy.
Whatever you think is going tosupport your needs the most, and
I think that's where the scaleis going to come from.
I think companies are going tosupport your needs the most and
and I think that's where thescale is going to come from I
think companies are going to befinding ways to package those up
and ship them off and make lotsand lots of money from them um
(31:51):
cognition labs is actually doingthat.
Speaker 2 (31:53):
They're a sas platform.
I guess it's like an agent as aservice as opposed to sas, but
like um, that's that's a gooduse case.
And then um always follow themoney.
You know, if you're you'relooking at the sequoias and the
vc um trends right now, theangel investors you know, agent
startups are attracting fundingum at a pretty pretty high rate
(32:14):
right now.
Um, so you know, check thoseout.
And then another thing too likesolopreneurs are going to be
able to really take advantage ofthis.
Just like, imagine like a, aone-person startup in richmond
that can have their own digitalworkforce.
And like fix stuff that needsissues like the water system.
Like for for those listeningfrom the other side of the world
(32:34):
, like at the start of 2025,richmond and ryko and the
surrounding areas had no waterright, what if there was an
individual that can get ahead ofstuff like that to anticipate,
you know, water droughts or whatended up happening with that
situation?
I think there was like amachine function messed up and
then it flooded somethingsomewhere and then it just
(32:55):
messed up everything right.
But like, I think, the AIagents and individuals with the
passion and the energy to kindof really, you know, take that
thing and run with it are goingto be able to do it by
themselves one person and thencome up with some great
solutions to really have apositive effect on the community
.
Yeah, definitely.
And just because we're here,will Milton shout out RVA Water.
(33:19):
You know he did a lot of goodstuff when we were short on
water in the richmond area.
So if you're ever in richmondand you see a cool little uh
aluminum bottle, that's one ofour uh founders.
Uh, out here they already.
Uh will milton um giving uhgiving water.
And great guy, great, greatcompany.
The water's way better than thestuff that you would get out of
aquafina or a designing.
(33:39):
So so you know.
A little.
Public service announcement.
Speaker 1 (33:43):
All right Go Will.
All right man.
Speaker 2 (33:51):
What do you think the next few years looks like for
AI agents, if the trajectory isgoing to go like the same hype
cycle with, like just AI ingeneral, I say what OpenAI
released, release chat, gpt inNovember of 22.
And then by the end of 23, forthe early adopters, I was in
conferences listening to AI andthen at the end of last year I
(34:14):
mean everybody is aware of it solike in two years, if it took
us to get there, I think AIagents probably get there in six
.
I think you know assistancewith you know agents is going to
start to really affect a coupleof industries.
So if you're in theseindustries, try to figure out
how you can show your employerthat you're using AI to help you
be more efficient.
(34:35):
So, marketing, coding,technology, all that, and then
customer support, I think I knowsome call centers are already
starting to introduce callcenter agents and things of that
nature.
So, like you know, make surethat you know.
You just kind of you know, geta little bit of knowledge
underneath your belt so you canbe ahead of the competition.
You know, because if they haveto cut staff, you might want to
(34:57):
cut the staff who has nounderstanding of what's going on
with AI agents versus theperson who's always talking
about hey, I saw this researcher.
I'm thinking this solution isbetter, right?
So I think, in about like sixmonths, what's the April?
Yeah, about the end of the yearit's going to be a lot more
plug and play, no code, easy toexecute and implement AI agents,
and I think it's just becausewhen you looked at where just
(35:19):
artificial intelligence came, ittook about like two years for
it to just be like.
I mean, like you can't goanywhere without anybody talking
about ai agents now, and Ithink that they've been talking
about ai agents for the earlyadopters at the end of last year
, for sure, and it's starting tokind of pop up a little bit
more.
So, six months, for sure, andthen long term, like five years
from now, the aging ecosystem isgoing to be like um, that's a
(35:40):
different conversation.
You know that we should have,but like, I think it's is going
to be like that's a differentconversation you know that we
should have, but like I thinkit's just going to be a
different use case for people towork.
I look forward to seeing how Ican compete in that market.
You know like I'm trying tostay as knowledgeable as
possible.
So I'm going to the event onTuesday and you know anytime
that if you're interested andwant to get a cup of coffee, you
know, check me out onSustainable creative.
You know, I love just havingthese conversations.
(36:01):
I love hopping on with jasonyou know it's early, he got, he
got to get the garage knockedout, I got to cut the grass.
But it's like whenever you taketime out to like, you know,
just be humble and haveconversations like this, you
just get better.
So I say, um, you know, sixmonths, it's going to be a
little bit quicker mass adoption, but in, like you know, five
years, um, it's, it's going tobe hard to predict.
I mean, differentadministration, different world,
(36:23):
it's going to be interesting.
Speaker 1 (36:25):
Yeah, maybe different administration.
I like it.
Look, I would love to have thatconversation.
We're not having thatconversation.
2028 is very unpredictable.
Speaker 2 (36:37):
Yeah, but.
Speaker 1 (36:38):
I do think that you're right on.
With the vast change that'scoming, it's hard to tell what
agent to agents are going to beable to do and what big tech is
going to do with these thingsand how they're going to mass
produce them and roll them outinto society.
Right, it's like you'reabsolutely right.
It's like do what you can todayto learn and grow with these
(37:01):
technologies, because there's nogetting around it.
They are in our world andthey're going to become more
prevalent, more available.
Get to know them and don't beafraid of them.
Right, don't be afraid of thetechnology and trying new things
with them and seeing if you canexplore different means of
leveraging them so that they'revery approachable in your work
(37:24):
and life.
And when your company decidesto start leveraging this stuff
internally, you're ready for it.
But we're here to help AI ReadyRVA.
That is our mission.
We want to support thiscommunity.
So let us know how we can Reachout to us in LinkedIn.
Reach out to our friends atSustainable Growth Creative.
(37:45):
Reach out to them and let themknow how they can help you.
That's their business model.
They want to bring people up tospeed.
We also have a really goodplatform that we're developing
at AIReadyRVAcom.
It's going to help educatefolks.
So become a member today and wewill be contributing in great
(38:07):
force to the growth and scalingeducational opportunities for
our cohorts and everyone, allour members.
So join us online and get toknow AI today, because the
future is not coming, it isalready here.
Thanks again, vic.
Speaker 2 (38:25):
Well, actually, can I close this For people who stay
to the end of this podcast, Igot something bonus.
Can I do a bonus thing realquick?
Yeah, do it.
Have you heard of Manus AI yet?
Speaker 1 (38:35):
Yes, yeah, it's a Chinese startup, right.
Speaker 2 (38:39):
Yes, it is Just like DeepSeek, and that's another
thing.
I think that the um you know,the chinese are um, really like
what they did with leveragingolder um.
They weren't even blackwellgpus from nvidia, what they
actually did, and it's super inthe weeds um, and I'm open if
somebody in the comments cancorrect me if I'm wrong but like
(39:01):
all those GPUs from NVIDIA,they use CUDA as like their
programming language.
Apparently, deepseek figuredout how to get around that and
that's how they were able to getthose chips to function so
effectively.
But saying that, to say thisthat Manus AI is just as
disruptive as DeepSeek.
If ChatGPT is an assistant,manus is a project manager that
(39:26):
actually does the project workfor you.
Check them out.
They're a state of the art GIAagent benchmark and all that is
is just like measuring success,like they benchmark all these
tools and like Manus AIapparently scored pretty high on
(39:48):
that.
Check them out.
And then, from BC's 75 million,what a $500 million evaluation.
So it's not like they're a toolor a company that is not
turning heads.
That's a pretty high evaluationand that's.
You know.
You follow the dollars.
You know I was able to use itto come up with I'm doing a
score three-part AI seriescoming up.
(40:12):
I should probably pump that, butI ain't got the dates yet.
But it's a great tool.
It's using a whole bunch ofdifferent AI agents at one time
and it helped me come up with agood curriculum and it took me
maybe like 10, 15 minutes to getin there.
And then I just you know, tookthat information, put it in
gamma and I got I got a hell ofa presentation right.
So if you're looking to try tosee like what's the example of
(40:34):
what an AI agent can do rightnow, in the moment you can
download Manus AI and they got abeta version.
But just be particular, becauseit's just like Deep Seek.
That stuff is going to be on aChinese server.
And if you don't want yourstuff on a Chinese server,
there's a multitude of reasonswhy you wouldn't.
I'm not going to get into them.
Just make sure that you're kindof particular about what you
put on there.
I'm asking it to help me witheducational information and the
(41:00):
stuff they give back is legit.
I'm telling you there's goingto be a Chinese Coca-Cola coming
out if you do that.
So don't do that right, becauseI mean like this is the end of
the podcast and, like I alwayslike to reward people who listen
to the end, give them some youknow little Easter eggs to kind
of go back and, you know,hopefully talk about the stuff
in the comments.
Speaker 1 (41:15):
Awesome Thanks.
So you have a wonderful rest ofyour weekend and I look forward
to catching up with you againsoon.
Speaker 2 (41:22):
Yes, sir.
Speaker 1 (41:26):
And thanks to our listeners for tuning in today.
If you or your company wouldlike to be featured in the
Inspire AI Richmond episode,please drop us a message.
Don't forget to like, share orfollow our content and stay up
to date on the latest events forAI Ready RBA.
Thank you again and see younext time.
Popular Podcasts
Amy Robach & T.J. Holmes present: Aubrey O’Day, Covering the Diddy Trial
Introducing… Aubrey O’Day Diddy’s former protege, television personality, platinum selling music artist, Danity Kane alum Aubrey O’Day joins veteran journalists Amy Robach and TJ Holmes to provide a unique perspective on the trial that has captivated the attention of the nation. Join them throughout the trial as they discuss, debate, and dissect every detail, every aspect of the proceedings. Aubrey will offer her opinions and expertise, as only she is qualified to do given her first-hand knowledge. From her days on Making the Band, as she emerged as the breakout star, the truth of the situation would be the opposite of the glitz and glamour. Listen throughout every minute of the trial, for this exclusive coverage. Amy Robach and TJ Holmes present Aubrey O’Day, Covering the Diddy Trial, an iHeartRadio podcast.
Betrayal: Season 4
Karoline Borega married a man of honor – a respected Colorado Springs Police officer. She knew there would be sacrifices to accommodate her husband’s career. But she had no idea that he was using his badge to fool everyone. This season, we expose a man who swore two sacred oaths—one to his badge, one to his bride—and broke them both. We follow Karoline as she questions everything she thought she knew about her partner of over 20 years. And make sure to check out Seasons 1-3 of Betrayal, along with Betrayal Weekly Season 1.
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.