Join us as we delve into the fascinating world of Artificial Intelligence (AI) by interviewing the brightest minds and exploring cutting-edge projects. From innovative ideas to groundbreaking individuals, we're here to uncover the latest developments and thought-provoking discussions in the AI space.
you need to ask the right question and oftentimes it's not enough with with one
question because one question might lead to five new questions and then you need to answer those right so there is a back
and forth reasoning that is happening which is actually very interesting because that's how our brain works so we
we had the luxury to to to build two companies previous to this one that were
successful and I have to say like we are two people but I believe that future if you look at the future and if you look
how many of these tools that we use are open source you know so I've never ran a
team bigger than I have now I would say probably half the work that's being done there in the corporate office could
probably be done by some kind of automated automated uh agent like like
super agent it just it's just the facts of the matter so and that's probably like that through a lot of corporate
America in some ways making life easier but it's also uh eliminating that sort
of uh autonomy in our lives that we're doing this sort of thing or do you think it's or do you look at it more as it's
bringing people up to do more important tasks
so my name is Ismail I'm one of the co-founders of super agent one of two
uh I basically write all of the code and
I'm the CTO of the company and the basically the initiator together with
Alan um on the open source project that is super agent basically so that's my
background is I'm a chemist weirdly enough but uh never graduated always
been building companies build been building companies with Alan for the past I don't know 12 13 years so we know
each other pretty well been through everything you could be as Founders which is a strength I guess so
that's that's my background basically and yes my name is Alan and I'm the
co-founder of super agent together with Ismail as mentioned I have an engineering background
mechanical engineering and uh we are from the city we are based in
Gothenburg Sweden which is the city of Volvo so over here like mechanical engineering is a big thing
so everyone works for Volvo here except me and Ismail you know we always wanted
to do something something different so we we've always been into
like uh cool and interesting technology bleeding edge stuff Leading Edge like we we've
always been into that uh and I think our strength has always been like you
know how can we toy around with cool technology and create like something
that is useful right and that I think is our kind of our Niche and oftentimes we end up in
like very boring Industries for some reason I don't know I think we're boring boring people deep down so we gravitate
towards that yeah yeah I mean I remember we did like serverless deployments of
like apps before like versel was a big thing uh when they were like you know a
couple of developers we were there like doing stuff in production uh so yeah we
would like the bleeding edge stuff awesome that's very cool so to get into super agent a little bit
uh for people who don't know or anything about super agent but are interested
what what is the project do you want to take the easy smile no I think you should take I think you should
take the part about why we are doing this like what the whole thing is for somebody who doesn't know what what
field we're in like AI agents I think it would be appropriate to just give a
proper background Yeah so basically the idea behind agents
uh within the field of AI is not something new right
um you know from the dawn of computing by Computing and AI was like invented at the same time right that's basically the
history of computing so and the idea of Agents is you know dates back to that
and the whole idea is that you have some kind of program in this context that can act
autonomously right can understand its environment and can basically on its own
solve different tasks right and reach different goals so like
that's like the theoretical background which we are leaning on here right
and what happened during the last year with this whole you know Transformer
Revolution that we're we're in the middle of is that people started playing around
with toys like chat GPT and the underlying
large language models and realized that an llm or a large language model can
actually serve excellently as like a reasoning engine
for an agent right so if you if you imagine an agent having all of these different components
um like a brain If you will that it needs to perform a certain task
the llm happened to perform excellently as like a reasoning
engine so people started creating different kind of hacks and different kind of
demos based on this and the results like were
very interesting to say the least right you you've seen the demos it's very
promising and Ismail and I were we were in in the
middle of building a product that's sort of related but also unrelated and
we started borrowing borrowing some ideas
from the community and we quickly realized that
if we would to use an agent to solve like a software problem that we had it
had a lot of different benefits right because we quickly realized that
there is a a paradigm shift going on so the old way of Building Products the
building software is basically the programmer has to embed knowledge into the product by
programming that's essentially what programming is but the new way is that you Outsource
all of that to the larger language model because a large language model already has a bunch of knowledge embedded into
it right so that's kind of that's what that was our aha moment we saw that okay
things are gonna change pretty fast looking on you know we started playing
around building our own agents and we quickly realized that okay all that existed out there was demos
there was no real way of building an agent that you could actually use in production
so that was kind of the the starting point of of super agents
so we're used to sort of building super agent for yourself in a way to make your life easier is that how it got started
exactly because we were we were building a different kind of product we were
building a product called clickable which is basically an AI driven ad
generator right so basically we we would generate
an image with with diffusion models and then we would generate like captions and
AD copy with large language models and we wrote so many lines of code to be
able to perform this quite simple idea the results were never
as good as we wanted them to be however as soon as we shifted to the agentic model
and leveraged the know-how embedded in the llm
the result went through the roof right yeah it wasn't that was a true
paradigm shift yeah and that was yeah that was like the uh basically the uh
beginning of of super agents so it was the back end for this AI assistant for
marketing teams basically and so just over a week uh I decided to you know try
to formalize this because if you could build a marketing assistant like a co-pilot for a marketing team what other
cool stuff wouldn't you be able to build and how cool would it be if that stuff could be ran in production in a real app
with real users using it right how cool wouldn't that be so that basically became the embryo of
super agent we decided to you know open source it because we believe in the open
source model for for building these type of uh Frameworks and Technologies most
of super agent is basically built on open source technology so we wanted to
contribute we wanted to create something that was super valuable for someone like
Ryan who is not like a machine learning engineer so we wanted to simplify you
know Ryan's life and enable him to basically bring in AI agents into this
application with as simple as like implementing stripe for payments basically that's like the the gist of it
and when were you guys working on this project for the marketing is that it sounds like you've been working on that for a while now more more time than we'd
like to admit I think uh close to a year now
but you so you did it before the AI hype was out there exactly yeah yeah so so
when the when chat GPT launched we already had like tens of thousands of
signups and then when chat GPT dropped like our traffic and our signups went through the
roof so this was before the hype um so we spent a lot of time on it and
that's like you know this just goes to prove like
how powerful the authentic approach is because we scrapped all of our code
and replaced replaced the the whole thing with a number of agents that
collaborate to bring the user customized ads right
and if you think about it it makes sense that's how an ad agency works right if
you remember that like the series Madman yeah don't rape run those guys that's
how they work you have a copywriter you have the production you have the art director they need to collaborate in
order to bring something to life and that's how our agents work in in clickable right
so what do you see is the biggest complication with your technology right now or what roadblocks do you see that
you have to overcome I think uh Tech wise
um there are there are two roadblocks I think uh for AI in general uh
one of them is that people if if people are going to adopt this technology let's
say if a law firm is going to adopt some some AI assistant or AI agent into their
into their daily operations the thing it needs to be is that it needs to always
be 99.9 correct at all times and so the
inherent problem with AI or large language models in specific is that it's not always 99
nine percent sometimes I I read some benchmarks some people have done it was
like 40 49 so it was the hit rate was is
really really low so there are ways of of bringing it to 99.9
and the that's the Frameworks surrounding Frameworks like super agent
or other Frameworks out there that are surrounding and enhancing the output
basically of of the language model so that's something that we need to work
even harder on and that's something that we are actively working on so we are training our own
algorithms basically to to handle this on our side and I think the second thing
is that you know when you write code the old way you always have control over issues and
bugs and stuff like this for the Developers for the tech team basically with AI agents it's a bit hard to trace
and actually see it's it's it's rarely a software bug it's an AI bug it's a
completely different type of bug that appears it's not that the like application breaks it's more like the
answer was incorrect the llm didn't didn't answer correctly
so it's not a bug but it's it's not the correct answer either so that's really hard as well that's something that needs
and these two go hand in hand of course so these need to be addressed and I think there's a lot of interesting
projects out there trying to address them yeah L M's are by Nature black
boxed right and that creates a lot of usability issues especially in like
business critical contexts there you need to have a lot of control
and you need to be perhaps perhaps you know adopt a different
approach than like the zero shot approach which is where everyone starts out right just
write the prompt and then hope for the best that's how we use chat Chipotle basically right but um
in order for them to to perform in the way we need them to perform you actually need to step out
outside of the llm and you need to do a lot of stuff without getting into detail
you need to do a lot of stuff [Music] before after and like
around between elements Yeah you sort of want it are you sort of you sort of need
to ask the MLA the llm the right question is what what you're saying like you need to ask the right question and
oftentimes it's not enough with with one question right because one question might lead to
five new questions and then you need to answer those right so there is a back and forth reasoning
that is happening which is actually very interesting because in a way that's
that's how our brain works right it's very seldom that we just think a thought
and then we Act and that's the end of it oftentimes you go back and forth and you have like some
tasks right sure uh so so we actually like we learn about ourselves a lot when
working on problems like these and and this is kind of an unexplored
territory as as Ismail mentioned sure what what so how
I mean this is a black hole but we'll start off here what L alums do you guys use right now for for super agent
so super agent supports hundreds of llms it supports any like it supports the big
ones of course like open a open AI Claude uh llama all of these big models
but it also supports whatever model that you publish uh open source or if you
publish it like on hugging face so you can use whatever llm you want so we have
people using specifically fine-tuned llms that they have trained on a certain
Corpus of data to do specific things um so we don't you know we are a
developer tool it's up to the developers to decide what language models they want
to power their agents with we just give them the tools to do that in a you know
production ready and developer friendly way okay so like say if
you know we Hunter and I we sort of before we do these interviews we don't want we don't like to research too much
about the projects because we like to hear the answers from you guys um yeah so one of the things like say if
I'm a developer and I want to go download and play with super agent um you they would someone would go on
GitHub download the code and then plug in what llm they want to use at that point uh actually we yeah that's one way
of doing it since it's open source they can actually like spin up their own uh thing and then and run that but we also
have like a hosted Cloud solution so you can run super agent by going to
app.superagent.sh and log in create an account create your agents like in a in
a nice user interface that we built for you and then try them out and if you're happy with them you can integrate them
into whatever application you're you're running basically so if for you guys you
do a lot of podcasts you want to you don't have time to research everything so perfect use case for you guys would
be to build a research agent so what that agent would do would be to you know
go and do all of the research for you for the project highlight five things and you you just pick up on those five
things and and you're ready to go you don't even need to do the extensive research and you can do that in super
agent in like one minute so that's the power of it and you could be a developer you could even be someone
like today I spoke to a designer that created agents for himself so he doesn't know anything about code
but he can do it with super agent and run it for his business so we have law firms doing it we have consultancy firms
doing it we have other technology companies using our platform and we have designers so it's a pretty broad scope
of people which is amazing yeah what we have found is that it's
oftentimes two different categories of people involved in these projects
first off you have like the prompt engineer or the the agent designer whatever you want to call them and
that's often someone in charge of a workflow oftentimes a business person or a product owner of some sort
he who understands the business requirements and goes in creates the
agent writes the prompts Etc and then tests it out and when it's okay
hands the agent and the agent's API endpoint because the agent gets its own
endpoint right so it can run from any app and that hands it over to the developers
and the developer doesn't have to know anything about AI or llms or anything like that
again the developer works with whatever app they're working on right and simply
implements this so that's how that's how we envision
the users using super agents okay so right now there's the the cloud-based
solution and the sort of more the techie nerd developer uh code on GitHub if
someone went on the you mentioned the designer you know uh used um super agent what does that flow look like in terms
of if someone wants to go and use super agent they go to the website and with the like for that Designer example what
did they actually do in the on the website to get what they wanted done yeah so in this specific case uh the
designer had some documents uh like design guidelines
brand guidelines and stuff that he wanted to basically answer you know
wanted to ask questions to those documents get answers and then act on those answers on behalf of his uh
customers or users or whatever it might be in his in his in his daily work really as a designer he wanted to
leverage leverage that data so what he does is he just logs in with his account
he he he basically goes into a configurator we have he presses a button
he chooses what agent he wants he chooses if it's you know should as a should have access to the internet if it
should have access to documents videos images whatever it might be right and so
we he just enables the access to to whatever data basically he wants to to use and then he can just press a button
and run that you know and he gets a chat interface if he wants to he can use it
in a in a more technical way if he wants to later on which he's planning to do
and uh so it's really simple for him and that's basically our approach is to
build something for people that don't necessarily have the skill set to build
this on their on their own okay that's that's like the uh
just to sum it up to actually build an agent there's no there's no coding
involved that's the beauty of it because the coding is already done that's the whole
idea so you just click around in the user interface and you configure the
agent by writing prompts that's it which is we just we interviewed Otto GPT last week and um
they've got a ton of hype and I think they're like they're they've got more favorites on GitHub than like react and
node.js yeah and it's a great tool but it's just a command line interface right
now um yeah it's crazy you guys talking about this because you can people can go out there and use this right now like
and they don't have to be developer which is awesome I feel like you guys are under under um
an unknown project in a way
yeah yeah and we I mean we collaborate with with a lot of these uh hyped or not
even non-hyped uh AI projects out there because what happens in the end is like
our users you know some of them want to run you know other type of Frameworks
such as Auto GPT or baby AGI two of the biggest and they can do that on super
agents so that's the cool part we try to include everyone like we don't
discriminate any any Frameworks or any other type of Agents we don't have any
real preferences on that we only want to serve the users so what does our users
you know what do they want to run do they want to run baby AGI we'll fix so you can run baby AGI basically that's
our approach to it uh so and yeah so that's pretty cool so we can collaborate
with anyone basically and we do a lot of collaborations so you guys seem to have a lot of different use cases do you guys
do any Data Tracking or do you know how many people are using super agent or
what kind of interesting use cases you've seen for it so far yeah we do we do Allen is the tracking
King what's the most common one too that you see so our tracking at the
moment is underdeveloped but we have um that's why he's referring to
me it's never going to be perfect yeah so we have uh over a thousand users
active users on our Cloud solution and we have a lot more using our open
source solution uh self-hosted and we have a bunch of different use
cases uh very the common ones because we're in the middle of the chat with
your document Trend right we see a lot of startups doing that we have a lot of that we have a lot of
assistance type use cases as well
so for example you have an assistant where you load a bunch of know-how of
some sort could be like your knowledge base for your company and you can also equip the agent with
some kind of capabilities like post to slack whatever it might be that's a very
common use case as well uh as well would you add an interesting
yeah I mean I mean in the end I've seen like because I speak to to users every
day we have a Discord server that has around I think soon 400 developers in
there and they bombard me with a bunch of like you know ideas and prompts I've
seen some crazy prompts I've seen some prompts it doesn't even look like a language it looks like I don't know what
it is just emojis everywhere you know and so I've seen it all out there I've Seen It All I've seen the dirty side
more efficient business yeah but what I would say is that the trend
is that people and I've thought about this myself and that's why I love super
Asian because I'm a user myself uh people want to replace themselves that's the weirdest thing people want to
replace you know all the tedious BS tasks that they do every day and some of
them are really important uh but there there's so many of them that you can't do everything you know perfect so in
that sense it's perfect to just create your own personal bot I've I've done it now so I have like a super bot that I'm
testing yeah exactly and so what it does is basically if you have questions about
how to like programmatically Implement super agent in your own app or something
like that you'll just ask it and it just writes all of the code for you you know so you don't have to even write the code
for it and it's and you can ask it anything like how do I do this how do I do that blah blah blah and it just
writes it you copy paste it and you're done you know that's that's what people want to build and that's that's pretty
amazing because regular individuals want to do it uh you
know I believe you know smaller businesses want to do it larger businesses want to do it that like
everyone wants to do that even the housewife in the end will want to do that uh so it would be pretty amazing to
build something for that housewife in the end you know anybody can do whatever they like replace themselves which would
be an amazing thing to do yeah so yeah so I wanted to ask you know
what are the limits to this is it just the is this just the human imagination
or uh what is there a specific limit where it's like okay it does these these
it creates these sort of Agents very well but there's these certain type that could be improved upon or once we start
stretching it to these purposes it becomes difficult uh where are the limits with that so yeah this is we had
this discussion this morning actually and because you have different kind of
you can design an agent using different kind of Loops we call them
and different kind of Loops for example the react Loop is one planets always
another they're good at different things
however to answer your question
the limits are actually kind of unexplored right so there has been
research on on what types of tasks each Loop is good
at right this is quantifiable there are different standardized tests that they
put these agents through um for example if you need something that
is more an agent that is more reasoning and can
do something that is not sequentially planned in advance perhaps you want a
react which is a reason an act Loop that does that because it can
continuously uh you know take in its environment you
know think about what to do Etc large language models does that very well
um but you might have other use cases that doesn't require that
and we had this discussion this morning and you know it's very difficult because we are in the in the just in the
beginning of this and the research is like very fresh the articles that are coming out so I don't think there is any
kind of answer on on what the limits are that's the first part of my answer
the second part of my answer is if you look at the scope which we have
discussed today the scope we have discussed today is quite narrow actually Ismail
um mentioned a basically a support version of himself right and he's got a
ton of Leverage from that agent basically a customer support agent right
and we see our users creating similar stuff as mentioned
that's kind of low-hanging fruit but if you look at the workforce out there if
you look at any like big company and you look at the average employee what are they really doing and how high
is that bar right it's not very high
so there's no I don't think you you need to reach the the limits of what is
possible with this kind of Technology before you can create a tunnel value
that's kind of the hypothesis we are working from so we are very happy where
we are right now when it comes to like output from these agents and that just goes to that's just that
tells you more I think about you know
what people actually are doing at large companies for example yeah well I worked uh like Corporate America for example I
worked at Sherwin-Williams they're a paint company here in Cleveland I probably shouldn't be calling this out but
um they're the largest I think they're one of the largest paint suppliers in the US um and I would say probably half the
work that's being done there in the corporate office could probably be done by some kind of automated automated uh
agent like like super agent it's just yeah it's just the facts of the matter so and that's probably like that through
a lot of corporate America yeah exactly that's you know that's how large organizations tend to and there's not
there's no I don't think there's any danger singling anyone out because
they're not the only ones so that's kind of the the hypothesis we are
basically working from now is the Super 8 is it
um is it completely free or is it paid or how does that work for the cloud version yeah so we are
planning to charge for it but we are not right now because we are in open beta
but we're planning on releasing a paid plan soon gotcha and can you guys talk about the
tech stack how you built this what it's built with um I know this could probably you could probably talk about this one for uh for
hours but if you can give us uh an insight to how this was built um you talked about a little bit the background
of it starting from a marketing clickable but what it what's what are
the tools you actually built the super agent with so uh the actual Tech stack it's a
pretty simple Tech stack actually it's not something Advanced it's basically a uh we uh we write all our code in Python
so we use a framework for that called Fast API amazing framework uh and what that does is that it
basically creates an API that any developer can use in to integrate into
our app and then we have you know different type of Frameworks that are
open source that you can use and like we use Lang chain which is a framework that
is you know blowing up it was really the first big agentic framework for
developers and so we we use that we use llama index which is another one we use
you know a bunch of different uh type of tools
and to power our agents such as like zapier and other type of you know
automation tools web search and uh you know all of that stuff we
have a pretty simple you know database that connects everything that's run through a service provider called
superbase who's like amazing yeah amazing product I love it and also built
open source partly so and that's great for us and uh you know
for the the llms basically you can you know we we connect to hugging face and
we connect to a bunch of different third-party services like notion HubSpot you can bring in whatever data you want
or whatever language model you want into the system so we have Vector stores databases a bunch of stuff but really
simple and the front end is next JS it's it's deployed on Versa we are big fans
of Versa I mean we've been using them since I don't know you know we've used them
forever basically before they even were uh this big and and always been fans of
those guys as well so that's it's pretty simple stack and we have a lot of contributors which I want to give a big
shout out to you know they come and they help us out open source and develop our our framework with with all of the
amazing contributors and and you know so we're really agnostic we want to
bring in what people want to run and uh so we we are pretty like nimble and fast
move fast like and and have simple code that's like it but everything is on
GitHub so I urge everyone to just go and take a look and try to run it yourselves and play around with it
all right awesome and you got just to be clear are you both you you two started
the company is there anyone else or no is it it's just you two and then
pretty much contributors on um GitHub or in Discord yeah it's the ninja Army we call them
they're like ninjas there are super agents you know awesome so that's so awesome you mentioned how you know
people are using this a lot to replace themselves or replace these menial tasks at least uh
do you see this is a sort of maybe uh a risk of AI or a direction that AI
is going uh where it's kind of like in some ways making life easier but it's
also uh eliminating that sort of uh you know that sort of autonomy in our
lives that we're doing this sort of thing or do you think it's or do you look at it more as it's freeing people up to do more important tasks
I can answer this one perhaps so uh Ismail framed it as replacing
themselves I wouldn't frame it like that I would say that for in ismail's example
the the Ismail agent that was the Super Ismail agent that was created is
actually not replacing Ismail it's actually allowing Ismail to to get more
leverage out of his time clone cloned version exactly so it's
basically a clone so that this man can create more value elsewhere
and that I think that our users see that right because in the end the agent the the
super super smile in this case is dependent on Ismail right because it feeds off of ishmael's
know-how and the same goes for the others so I
think in the like this is a very philosophical discussion I guess but where we are right now like the agents
are very dependent on us humans for knowledge and for technology as well
well that will be over quite soon I think the latter one
very cool yeah I think that's that's a great way to put it for sure so what is your next direct what's your
direction for this project right now what are the next steps that you're currently working on yeah we're working
on a completely new version of uh super agent
that is going to make it much easier to create more complicated agents that can
perform much more complex complex tasks right because the agents today are quite
simple if you compare them to you know like our vision so we have a vision of creating agents
that can replace like 90 of our workflows in Corporate America
that's that's kind of what we are gunning for which is not a high bar at all as we were discussing earlier and
that's what our next version aims to do maybe you guys are slowly going to take
over Volvo I mean you're right by that live right by them you can take over that working that's too easy that's too
easy I have friends working here yeah they tell me the craziest things
yeah most of our users are actually in in the US they're not uh from Sweden I
mean Sweden is not really a big place and it's really small and and there are
some users but all of them are basically us-based some of them are even Chinese I
think I mean we have people from all over the world no one in no one in Sweden knows what we do we have friends
and stuff they never talk work with us no they don't know what we do
it's just not you know it's not like San Francisco or anything where tech people
it's not that Vibe at all it's an old like you know it's like an old industry
uh city worker City you know it's a lot of beer and burgers not much yeah you
know AI in our city so do you uh if you don't want to disclose this that's fine
are you guys looking to raise VC Capital have you already are have you raised any money already or you sort of just
self-funding this project we have not raised anything yet currently we are self-funding this project
um but our plan is to raise Capital quite soon so we are working on you know
we want to do our homework before we start raising so that's what we are working on diligently
we're working on the product we're working on the the right kind of user success cases
and then based off that we want to do our first round so is it just you guys working on the
project right now or is it do you guys have a team going on or it's the two of us yeah the company yeah and then we
have the contributors of course yeah and how did you guys yeah we met during our University days
this was a long time ago 15 years ago we ran a math club together
okay so that was the background and we were helping kids out in
like um in the suburbs over here so there's like
a segregation situation that's going on in Sweden and especially in our city so
basically we wanted to do something good we want to give back right to to our our
favorite City so we organized a student-run math club where we engaged thousands of people in
the end and we we managed to
[Music] Inspire and teach our favorite subject
which is math so that's where we met and we always like we always Vibe because
it was quite clear that our Ambitions were different than our peers in our math club we always
wanted to to build companies and we wanted to work on cool stuff and we
wanted to build like something big want something that could really have an
impact so we always had like different ideas and then we went on to work like
we got our first jobs and that didn't last long right after a year or two it's
like okay this is not for us and uh we just decided to start a
company and that's what we did and we've we've worked together since awesome are
you are you guys doing this full time right now or do you still have a job somewhere else or is this your mom no it's uh it's full-time full-time so we
we had the luxury to to to build two companies previous to this one that were
successful uh so we have the luxury of being you know working on this full-time and I
have to say like we are two people but I believe that future if you look at the future and and if you look how many of
these tools that we use are open source you know so I've never ran a team bigger
than I have now and and and it's not it's because there's so many contributors that you know that's third
30 developers basically contributing at a regular basis it becomes like 30 they're not employees but you still have
to take care of them you know you still have to review their code and do all of the stuff you do but you don't pay them
that's the only difference you pay them with access basically so uh so it's a quite large team
uh but we are the only Founders that's that's basically it I mean yeah this is
like the open source business model like how do you so it's it's of course like legally separate so the the actual
repository is its legally its own thing with its own license Etc and uh then we need to
create like a viable company on top of that somehow right that becomes the next
step and but now we are fully focused on the open source project because that's going
to be the basis of whatever comes next right yeah no it's it's uh fascinating open source is awesome like to have in
your case 30 developers that would cost you millions of dollars you know if they if it wasn't open source to pay them if
they were on your payroll so it's awesome you have a great great team of uh contributors
um yeah you mentioned before that the city you're like Sweden doesn't have a big Tech scene how did you guys get
into doing Tech I know it was through math but are there other people like you in Sweden is there a is there a city
where there's a tech industry in Sweden or is this well there is there is like a text Scene here okay so we have like
companies coming out of here like Spotify Swedish so there is and there is a huge interest
and we have like very good technical universities that produce a lot of good people but it's not comparable with like
Silicon Valley yeah sure it's it's a numbers game right so even in the US you need to move to Silicon Valley or one of
your hubs it's the same situation over here okay so um
so so that's kind of but we are planning on like our long-term plan is to to
because you know if you want to grow this company you have to you have to take it to to Silicon Valley
so that's basically our our long-term plan
okay well if you ever come over here we'll have to grab a beer or something yeah we would love that yeah we're not
like we are not geographically bound you know it's not like the old days right so we're not like geographically bound here
with like hundreds of employees right and we it's quite in the early days still so
sure it's doable cool it's kind of the way the world is moving right I mean yeah you don't go to
the office as much anymore and I think that the pandemic had a lot to do with that and now technology has a lot to do
with that for sure yeah for sure in our case we need access to the right kind of
people right so we will go wherever these people are and they happen to exist in you know a couple of hot spots
but of course Silicon Valley is one of them so that's the reason of why we would like incorporate the company in the
future this is like a tactical financing thing I guess but it's a lot of has a lot of do to do with
with access to the right people if you're starting an automotive business I think this is a great place
it's a hot spot for that you have you have so many people you can recruit right but we are not in that business
right right so do you are your goal is to to get to
Silicon Valley one day and sort of build the business from there from what you're saying that's the dream that's the dream we should have done we should have done
it long time ago yeah yeah but now Now's the Time
cool well I think as long as you take care as long as you keep building super aging you guys are on a great great path
so for funding for growing the business for everything yeah
we believe in that you know in what we're doing and we believe in the vision and uh I mean with the world uh if
anybody could deploy these AI tools for themselves that would be like amazing for Humanity I think
so uh you know small mundane tasks at volume at scale is just something
fantastic what do you guys think what do you guys think can you see a future where
we don't have this type of Agents running around in our virtual
environments doing things for us can you see that kind of future I'll let you answer first honor
at this point no after all the research that we've done the people that we've talked to even over the past few months
it's it's inevitable that it's going to happen it's just
um a matter of time how well these things are going to develop uh how fast they grow into a larger scale
um people adopting them I think is going to be the biggest obstacle I think the
technology will get there far before the people will get there uh invite people I mean like the mass public because I
think there's a lot of people who are reluctant to this this sort of thing they think that it's
you know kind of scammy or techy or something like that or they won't be able to understand it but as soon as it
becomes more applicable to people easier to use uh becomes uh more normal I think that
there's no doubt that this is going to be the direction that the things go with AI
I have a different take on that I think it will be adopted long before people are comfortable with them
and the reason is profitability right because we are in a situation right now
which is very strange first of all we have this AI Revolution that's going on
at the same time we have a
increasing Capital costs Etc interest rates going up Etc companies being more
and more squeezed to perform well in terms of profit right what's the first what's the easiest way
to to increase profitability to get rid of people right and replace them with something else software agents Etc so I
think this bottom line driven adoption is going to come from you know top down private Equity
firms Etc that's going to be like the the early adopters of Corporate America
we already see that happening too you know we're seeing that happening all the time I mean if you look at uh like the
whole Netflix news that just came out uh yesterday we're talking about they're hiring you know directors for AI
you know that in in they're paying them almost a million dollars they're actors
and writers are only getting paid like thirty thousand dollars a year you know they're willing to invest so much into
this AI because they know in the long term that's the best way that they're going to end up you know making money
and cutting costs and increasing quality as well uh there you go so they've
already the movie industry for example they've already started to deploy agents basically that's what you're saying so
they replace the yeah the actors with agents how many how many movies are on
Netflix do you know that like how many how What's the total number of movies sorry great question I don't know I'll
find out let's say what what do you think Alan you're uh you're uh you're the engineer here what what math would
you apply to just give a bench I have a friend who have seen all of them
no I'm not sure how many it's different from different countries right yeah but
if you get all of it all all countries everything movies and I mean video video how much
video do they have how much video do they have that's a different question yeah yeah because if you think about if
they would take yeah if they would take a model and they're just trying to ask yeah there we go
and if if they would if they would want to create writers they could do do that
I'm 100 sure they could do that with a with a you know they could they could replace
writers there are you know that wouldn't be a big issue for them technology wise
that's like two years away and why wouldn't you if you're if you
like the whole for a a company like Netflix that is listed on the stock exchange in the US they have an
obligation to their stockholders to make the most profit they can like that's one of their defining things
so if you have to fire people and hire do do something with AI it's sort of like you have to do it because you have
to look out for the the stockholders um another thing that's kind of interesting is we for our email we get
um like responses to our email all the time and or on Twitter Twitter as well I've
got a Twitter account for our email uh newsletter and a lot of times people they just crap and I I they hate it so
much but a few times like the same person that hates AI will ask like hey how can I use this in my business like
the guy yesterday he's a chiropractor and he wanted to know how to how to use AI in his business to make it more
profitable so I guess my point is like you can hate Ai and be afraid of it as much as you want but as soon as someone
shows you a tool that can make your business more profitable I think that's going to change a lot of people's minds
really fast because that's the that's the bottom line you know especially if you have a family business that's
exactly my point as well like the the bottom line driven adoption is Gonna Come long before people are comfortable
with it and that's actually it could be a problem right if people don't really think about the ethical concerns Etc and
just go for profit that could cause other issues of course yeah
okay according to my manual
imitation of an capable agent I have calculated that we have 20
000 hours of video content to
2021 years numbers
yeah but yeah and there you go there you go yeah thank you yeah they could they
could reproduce that number of hours in you know every day right yes
generated with with AI yeah I was watching um this came out like a
couple years ago have you guys ever seen that deep fake of Tom um Tom Cruise
yeah yeah I mean it's it's incredible how real it looks so yeah
if they can do a deep fake at Tom Tom Cruise what they can just make deep fakes of whoever they want to act in any
movie they want I mean yeah it's voice you know usually voices decoupled from
the movement of the movie right they put they sample The Voice up after and do do a bunch of that kind of stuff voice you
can replicate voice one to one I mean you won't notice that's that's already done that's there are commercial models
I can go take some some of your videos and replicate your voice hundred percent today in like five minutes yeah it's
scary yeah the next step is you know they're gonna
generate their own movie stars right I don't need Tom Cruise right yeah
they will make their own Tom Cruise which they own all the rights to and they make them super popular they own
the brand they own everything um Jesus that sounds No but that's the
natural Next Step yeah virtualization
Murray on the tech to stream you on the tech to produce the movies and you own
all the intellectual assets for like uh because you created the actors right
yeah so the actors like all the actors are keep saying like what about our rights but I think it's like with any
technology right they're they're going to get pushed out at some point and they have to yep Embrace AI or do something because I
think a lot of their jobs are going to be threatened but I could be wrong yeah this happens all the time in all kinds
of Industries and I don't think like like movie like actors have any kind of special rights to keep their jobs
however they might be societal impacts I don't think from movie stars going out
of business but the famous example is like if all trackers go out of business or are replaced that's gonna have like
societal impact because you will have a lot of unemployed male you know certain
demographic that can cause a lot of political Ruckus right and then you'll have to invent other ways of
solving that for example taxation Universal basic income etc etc and I
think that discussion is do to say the least right
yeah I think it's going to be whatever it however it plays out it's going to be fascinating over the next two to three
to five years to see what happens you know because nobody knows but it's going to be it's going to be a crazy ride it
feels like yeah no one is safe yeah
yeah we are well we are yeah yeah that's we are the ones gonna that's gonna be
replaced first I think me and Alan will replace ourselves
you're super you're super easy male's gonna take over Ishmael yeah that would be amazing
though so what what is what is your Discord that people could go to so it's called super agent I can send
you the link to the Discord if you want to post it together with a video but it's basically a super agent and you can
just go in there and and join and you know the links are on our webpage as
well superagent.sh and you can check out check us out on GitHub as well so you
have all the links to everything there and you can try it out run it do stuff with it replace yourself or whatever
uh uh so uh yeah one thing we want to ask from your
viewers and or your readers we want to hear about
the kind of things they want to do with agents the
kind of ideas they have for their life and for their business so the chiropractor you told us about
earlier that's awesome that's what gets us going right when I want to hear all
of these quirky things quirky ideas because that's in the end
that's how we're going to help people so please check out our Discord
post any kind of ideas or questions you might have we are very happy to answer them
perfect yeah we love love to to hear these ideas I mean I'm intrigued by it
to here I always love to hear the crazy ideas personally the whacked out ones that I wouldn't think about
uh that's how it starts right yeah and that's how you said uh that's a great
way to test the limits um and continue to improve the project for sure yep cool thank you guys but
yeah yeah thank you so much we really appreciate it we'll probably uh maybe we'll do a follow-up
um podcast in a little later but uh thanks so much for coming coming out and yeah
and then uh yeah for any of the viewers or uh Ryan and I's newsletter is fry
hyphenai.com we provide a weekly or weekday news and then
uh along with some tools and different stories and updates and a fun way and an easy to read way the fry AI the fry is
just for fun because Ryan's daughter likes french fries so that's how we got the fry we dig it
yeah so we do that and then on Sundays we do the Deep dive articles so we'll be doing one uh here on super agent
um along with the the video follow-up so be sure to check them out Discord and
their project it is super agent dot sh all right we appreciate you guys coming
on and uh it was just really nice talking to you guys and we definitely learned a lot we're excited about your
project oh thank you so much thank you so much for having us we really enjoyed it all
right thanks a lot guys awesome thank you bye