Everyone Gets A Software Company — Benjamin Guo, Zo Computer
Read the talk
Everyone Gets a Software Company: Benjamin Guo’s Personal Cloud
Zo Computer combines a cloud server, AI and hosted services in one workspace. Benjamin Guo explains how that arrangement can replace a fragmented software stack, then asks who should own the agents and intelligence built on top of it.
From a talk by Benjamin Guo
At a glance
Ideas worth remembering
Zo’s core architecture puts files, AI and hosted websites or APIs in one server environment. Its simple workspace is paired with root access, SSH and API or MCP control for users who need direct access.
Anthia’s example teaches a concrete workflow: send an inquiry notification, let the owner call promptly and generate a payment link. The reported $100,000 trajectory has no stated revenue period or independent verification.
Removing friction has tradeoffs. Direct live publishing hides deployment work, and browser purchasing makes buying easier, but the talk does not explain release safeguards or transaction approval controls.
Guo extends his critique of SaaS lock-in to cloud agents: users should ask who controls the environment and who benefits as the agent improves. His proposed publishing platform would retain those improvements for the agent’s owner, but the learning mechanism and ownership terms remain unspecified.
A computer that builds and hosts
Benjamin Guo opens with a small example of the computing environment he is proposing. While waiting to present, he says he built a website collecting the session’s speakers, useful links, profile pages, blogs and research about their recent thoughts. He offers a QR code to the page and $100 in AI credits. The website serves as his first example of using a personal cloud computer to create something immediately useful.
Guo introduces Zo Computer as one instance of a broader category: personal cloud agents. His background connects consumer software with infrastructure businesses. He started on the early Venmo team in 2013, met his cofounder Rob there and later became Stripe’s 80th engineer. Rob went on to become Substack’s first engineer. Guo’s presentation, including his computer costume, centers on a question that extends beyond those credentials: what would make a computer feel like a person’s own environment again?
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Why computers feel fragmented
Guo uses the Finder icon to describe his design ideal. He attributes it to Susan Kare and interprets its white and blue faces as the human and machine living in happy harmony. For him, the aim of an AI tool is to restore that feeling of being at home with a machine: people enjoy tools when the tools extend what they can do and feel natural to use.
His diagnosis begins with the distance between a first PC or personal website and today’s collection of applications, sites and services. Each service adds another place to navigate. Guo also objects to rising prices tied to AI features people did not request. He treats confusion, frustration and fear around AI as understandable responses to this experience, and cites vandalized AI advertisements around his company’s Bushwick neighborhood as an example of that hostility. These observations establish his motivation; they do not measure how common those attitudes are.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Rent, lock-in and a home on the internet
Guo calls the underlying structure technofeudalism. In his analogy, users pay rent to SaaS providers, those providers pay cloud providers, and cloud providers pay the suppliers at the top of the infrastructure chain. The economic ladder matters because the user’s work and data remain inside products controlled by other businesses. Paying for access does not necessarily give the user control over how the product changes or how easily its data can move elsewhere.
He frames product deterioration as an incentive problem. A provider’s product manager is rewarded for improving the business’s bottom line, monetizing attention and retaining data. Those goals can diverge from making a particular customer’s life easier. Zo’s proposed response is to give people a home on the internet where their tools and information can live together. This is Guo’s argument about the structure of software businesses, rather than evidence that every provider behaves the same way.
Charlotte supplies the first customer example. Guo says Zo has been available for about a year and describes her as an LA-based private chef and life coach who hosts multiple websites on it. The same environment manages invoices, accounts, bookkeeping, scheduling and notes. The practical value is consolidation: the public websites and the internal work of running the business share one place. Guo reports that this makes Charlotte feel clear, calm and in control, but gives no quantitative comparison of her costs or time savings.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
The personal cloud as a server and workspace
Guo defines a personal cloud as a place that contains a person’s files, AI and hosted services. Because it is a server in the cloud, it can serve websites and APIs as well as hold private material. The important architectural move is to put the agent near the things it works on and the services it helps create. Guo describes those services as belonging to the user, although this explanation does not establish contractual ownership or the terms of the underlying infrastructure.
That cloud can also connect to local devices. Guo describes his own setup as a cloud computer containing his agent, another agent and numerous websites and APIs. It can communicate with his laptop or Mac mini, and he can use it from his phone. In this account, a personal cloud is the collective environment across those devices and the server, rather than a requirement that every task run remotely. He does not explain the protocols or permissions used for communication between them.
For publishing, Guo favors the immediacy of an older web workflow: copying files onto a server with FTP updated the live website. He argues that a nontechnical customer such as Charlotte should not have to think about deployment. This choice reduces the number of concepts between an instruction and a published result. It also puts changes directly into the live environment; the talk does not describe testing, rollback or other controls around those changes. His broader goal is one place for personal information, AI and hosted work instead of constant movement between services.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Anthia’s retreat business: inquiry, call and payment
Anthia, a nontechnical free-diving instructor and early user, illustrates a more complete replacement of a business software stack. Guo says she canceled subscriptions including Squarespace and Calendly and moved their functions to Zo. She hosts a retreat website on a custom domain, has several other websites and maintains a personal workspace reflecting her interests in fairies, drawing and mushrooms. The same environment supports both business operations and personal expression.
The central business mechanism is a short sequence connecting customer interest to a human response. When someone expresses interest in a retreat, Zo texts Anthia their phone number. She calls immediately, while the person is considering buying. She then asks Zo by text for a payment link and closes the sale. The agent carries information and produces the next useful action; Anthia still conducts the conversation and makes the sale. Guo attributes increased bookings to this faster response, rather than to the website alone.
Guo says Anthia is on track to make $100,000 and that demand has made her retreats difficult to book. These are reported customer outcomes: he supplies neither a revenue period nor a comparison that separates the software’s effect from other causes. The example nevertheless explains a concrete operational improvement—reducing the delay between an inquiry, a call and access to payment.
Behind the websites, Anthia has a retreat database, notes and accounting in one place. Guo emphasizes that she can benefit from a database without knowing what a database is. The abstraction hides the technical vocabulary while preserving the function. The same proximity helps with publishing: when she wants retreat images on a website, she can ask Zo to put them there, and the files can be hosted directly from her personal server. Guo calls this self-hosting; he does not detail how the database, accounting or payment service is implemented.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A simple interface with full server access
Guo describes Zo as accessible to parents and other people without technical backgrounds. He adds examples from businesses around his own company: a caterer uses it to manage restaurant staff, a recruiter is rebuilding his agency’s software stack on it, and a marketing agency has adopted it. These examples broaden the intended audience, but remain anecdotes rather than a measure of how reliably any new user can build a complete system.
For technical users, the interface sits over a Linux virtual machine with root access. Guo says users can host their own personal agents, bring existing AI subscriptions, connect through SSH and control Zo through an API or MCP. Root and SSH access provide a route to managing the server directly; API and MCP access provide routes for software and agents to control it. The product’s promise is that networking is already arranged, so users can start with the workspace and reach deeper capabilities when needed. Guo does not describe the networking configuration or access-control model before moving into the demo.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
The workspace: files, scheduled tasks and services
The workspace tour starts with model chat and files. Guo says users can choose models, bring an API key and interact with Claude Code inside the environment. A file system supplies cloud storage that can be used like Dropbox, while chat lets an agent work with those files. This joins storage with an interface for acting on stored material, rather than requiring the user to move files into a separate AI application for each task.
Automations add scheduled AI tasks. Built-in integrations and a skill library provide additional capabilities, while a browser allows users to log into websites and have Zo act there. Guo’s personal example is texting Zo to buy things on Amazon. He also identifies a behavioral tradeoff: removing the friction from a purchase makes impulse buying easier. The tour does not explain approval boundaries, credential handling or what happens when a scheduled task or browser action fails.
Hosting extends beyond a personal webpage. Guo says the server can host arbitrary HTTP or TCP services, and Zo also supplies a personal website with live editing through chat. His scheduling application illustrates why custom software can matter even when a standard SaaS product exists. It replaces Calendly with a setup he prefers, including agent review of people requesting time. He does not want unrestricted direct booking, so the application preserves a screening step rather than merely copying a conventional booking page.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Whose cloud, and whose intelligence?
Guo’s closing argument expands from personal workspaces to the structure of the internet. He wants individuals to have a durable presence comparable to the presence companies already maintain, with more direct interaction between people and between people, companies and their AIs. He predicts that agents will become a primary interface, acting on websites and APIs on behalf of users.
That prediction creates a second ownership question. Some agents may run locally, but Guo expects many to run in the cloud. Once both the agent and the services it uses are remote, the location alone says little about whose interests the arrangement serves. He asks who controls that cloud. A convenient agent interface can still leave a person or company dependent on a provider.
He discusses a company-wide cloud agent that holds company context and can be used by multiple people. Guo finds that model ergonomic, but argues that improving a provider-controlled setup may strengthen the provider rather than create intelligence the company owns. He calls this intelligence feudalism: accumulated value moves upward through the same dependency structure. This is an argument about control and who benefits from improvements; he does not establish a specific training, data-transfer or learning mechanism for the provider he criticizes.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Publishing agents that improve for their owners
Guo proposes that individuals and companies should be able to publish agents, own them and serve end users through them. In his desired arrangement, usage improves the publisher’s own agent. This adds a feedback loop to the personal-server idea: ownership should extend to the capability that develops as people use a service, as well as to its files and running software. The talk does not specify whether that improvement comes from memory, revised instructions, model training or another mechanism.
He says Zo has a beta version of this agent-publishing platform and closes by inviting people to sign up, contact him and obtain AI credits through his personal space. The page also contains the session’s speaker information. The closing proposal therefore goes beyond helping someone assemble a private business workspace: it aims to let that person publish an agent for others and retain the benefits of its use. The publishing and improvement system is presented as a beta, without a demonstration of its feedback mechanism.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Read the complete timestamped transcript
- 0:01
[music]
- 0:12
>> Hello everybody. I'm Ben from Zoho
- 0:14
Computer. I just posted on X from my
- 0:16
Zoho.
- 0:18
You know, I'll be giving away $100 in AI
- 0:19
credits and I do all sorts of stuff with
- 0:21
my Zoho.
- 0:23
This is my computer. My Zoho is my
- 0:25
computer. And before I get started, I
- 0:27
made the site just now while I was
- 0:29
waiting to get started presenting. This
- 0:31
is a kind of overview of all the
- 0:33
speakers in today's claw session. You
- 0:35
can scan this QR code and check it out.
- 0:37
I've got some useful links in here, but
- 0:39
also I've got like some deep research on
- 0:41
everybody presenting including their
- 0:42
recent thoughts, links to their profiles
- 0:45
and
- 0:46
websites, blogs, etc. So, scan this.
- 0:49
It's super useful. I just made it just
- 0:51
now. This is my computer that you see
- 0:53
right here, my Zoho computer. I'll be
- 0:54
talking about it and this talk is about
- 0:56
kind of Zoho and a lot of things. It's
- 0:59
also about kind of personal agents,
- 1:00
personal cloud agents in particular,
- 1:03
which Zoho is one instance of and kind
- 1:05
of the future and where I see all of
- 1:07
this going.
- 1:09
So,
- 1:11
a little bit about me.
- 1:13
I'm the co-founder of Zoho. My name is
- 1:15
Ben. I started my career on the early
- 1:17
Venmo team back in 2013 and um
- 1:20
I met my co-founder Rob at Venmo and I
- 1:23
left Venmo to work at Stripe quite
- 1:25
early. I was one of the first I was the
- 1:26
80th engineer and my co-founder Rob, he
- 1:29
went on to build Substack. He was the
- 1:31
first engineer there. And a final fact
- 1:33
about me as you can see is that I really
- 1:35
love computers. I'm dressed up as a
- 1:37
computer.
- 1:39
And I like to start with this like fun
- 1:41
fact about this finder icon. So, this
- 1:44
icon was designed by Susan Kare, an
- 1:46
early graphic designer at Apple Computer
- 1:48
and many people don't know the story
- 1:50
behind this icon, but it actually
- 1:52
represents the human, this white face,
- 1:55
and the machine, this blue face, and
- 1:57
they're kind of like in happy harmony.
- 2:00
And I think that is what we should
- 2:01
aspire to as we build AI tools for
- 2:04
people. Like humans love tools, humans
- 2:06
have always loved machines, and like
- 2:08
when we feel at one with our machine, we
- 2:10
feel like this. We feel happy.
- 2:13
So, who like me misses the way that
- 2:16
computers used to feel? Raise your hand,
- 2:18
right? Like the early internet, like
- 2:20
your first website, your first PC.
- 2:22
Right? It used to feel different, right?
- 2:26
Today,
- 2:27
the internet and our computers feels
- 2:29
more like this. We are swimming in the
- 2:32
sea of applications and sites and
- 2:35
services and
- 2:36
it's not great.
- 2:38
And on top of all of that,
- 2:40
they're jacking up the prices on these
- 2:42
things that we use, often because of AI
- 2:44
features that we didn't ask for.
- 2:48
And
- 2:49
I think quite reasonably, consumers or
- 2:51
I'll just call them regular people are
- 2:53
upset or confused or scared of AI.
- 2:57
So, Computer is based in Brooklyn, we're
- 2:59
in Bushwick actually, kind of like the
- 3:01
heart of hipsters in New York, and you
- 3:04
know,
- 3:06
there's a lot of this out there. I see a
- 3:07
lot of this kind of vandalism over AI
- 3:09
ads in the city.
- 3:12
On top of all of that, this is what our
- 3:14
industry looks like.
- 3:17
But the deeper root of why
- 3:20
this is bad is is a structural thing.
- 3:23
It's what we call techno-feudalism.
- 3:26
So, basically, feudalism was this like
- 3:28
kind of shitty system back in the day,
- 3:30
and though it no longer exists, it still
- 3:33
is alive and well in our digital lives.
- 3:36
So, we are still the peasants, we pay
- 3:38
rent to the SaaS providers who pay rent
- 3:41
to the cloud providers who pay rent to
- 3:42
the kings who are, you know, like
- 3:44
probably Nvidia these days.
- 3:47
And because we are peasants, our lives
- 3:49
are shitty. We are fragmented between
- 3:51
all these different tools. There are
- 3:53
various services that we use that lock
- 3:55
us in and and shitify over time. Like
- 3:58
that PM at that product that you use is
- 4:00
not incentivized to make your life
- 4:02
better. They're incentivized to improve
- 4:04
their bottom line, to monetize your
- 4:06
attention, to take your data, lock it
- 4:08
in, and sell it back to you. And because
- 4:10
you don't own anything, you are a
- 4:12
peasant.
- 4:15
So Zo, our mission has always been to
- 4:17
solve this by giving people a real home
- 4:20
on the internet.
- 4:22
And we believe that most people in the
- 4:24
world are kind of living in the Matrix,
- 4:26
and we're trying to show people
- 4:28
a better way.
- 4:31
So Zo is alive and well. We've been out
- 4:34
for about a year, and real people are
- 4:36
using it. Real regular people like
- 4:38
Charlotte, this LA-based private chef
- 4:40
and life coach. She hosts her websites,
- 4:43
she has multiple websites now, on Zo. Zo
- 4:46
manages her invoices and her accounts
- 4:49
and bookkeeping and her scheduling and
- 4:51
her notes. And it's just like amazing
- 4:54
system. It makes her feel like clear and
- 4:56
calm and on top of the world, and it's
- 4:58
fun. As you can see this text
- 4:59
that she sent me.
- 5:02
So what is Zo? Zo is a personal cloud.
- 5:06
We've never really had a personal cloud
- 5:07
before, but it's quite simple. You have
- 5:10
your own home in the cloud. You're not
- 5:12
using other cloud services. You have
- 5:14
your own thing. You put your stuff in
- 5:16
it. You put your AI in it. And because
- 5:19
it is
- 5:20
a place in the cloud, you can host stuff
- 5:21
in it. You can host services like
- 5:23
websites or APIs, and they're all yours.
- 5:29
And I think for people in this room, if
- 5:31
you're like interested in clouds and
- 5:32
personal agents, like your personal
- 5:34
cloud might look something more like
- 5:35
this. You might have like a suite of
- 5:36
local devices. Um and for me, I have
- 5:39
like this like Zo computer, this like
- 5:41
cloud computer as well, which has my Zo
- 5:44
agent on it. It also has a Hermes on it,
- 5:47
and it has my services. I have many
- 5:49
websites and APIs and things that I've
- 5:51
built, all hosted in the same place, and
- 5:53
it can also talk to my laptop or my Mac
- 5:56
mini, or I can use it on my phone.
- 5:59
And that, collectively, is my personal
- 6:01
cloud.
- 6:04
So, Zoe is a personal server with AI
- 6:08
built in. It's It's quite simple.
- 6:10
And it hearkens back to a simpler time.
- 6:13
So, back in the day, this is actually
- 6:15
how we built and hosted websites. We
- 6:17
just FTP'd our files onto a server, and
- 6:21
that updated our website. You were just
- 6:23
shipping to prod. You're like always
- 6:25
doing it live. And that's actually, I
- 6:27
think, how most people should build. I
- 6:30
don't think it's necessary for a regular
- 6:32
person like Charlotte to think about
- 6:33
deployment.
- 6:36
Unfortunately, for many of us in this
- 6:38
room, this is probably what your life
- 6:40
looks like, actually. You're kind of
- 6:43
just navigating between these different
- 6:45
AIs and devices and SaaS services, and
- 6:48
it's very confusing.
- 6:51
Zoe is meant to be a simpler way. It's
- 6:53
one place for all of your stuff and your
- 6:56
AI and the stuff you host, all of your
- 6:59
personal stuff in the cloud.
- 7:03
Here's another case study that I'd like
- 7:04
to talk about. This is Anthia. She was
- 7:06
one of our early users. She's totally
- 7:08
non-technical. She's a free-diving
- 7:09
instructor, and she's already on track
- 7:11
to make $100,000 on Zoe. She used to use
- 7:15
all of these different SaaS services to
- 7:18
run her business and her life, like
- 7:20
Squarespace, like Calendly, etc. And now
- 7:23
she's replaced all of that with Zoe.
- 7:25
She's canceled all of those SaaS
- 7:27
subscriptions. She is no longer a
- 7:29
peasant.
- 7:31
Anthia went from this,
- 7:33
kind of being very confused and having
- 7:36
her data in all of these different SaaS
- 7:38
silos, and not really being able to
- 7:39
connect them very easily,
- 7:42
to this. So, Anthea not only hosts her
- 7:45
retreat website on Zo on a custom
- 7:47
domain. She has many different websites
- 7:49
now. She has her personal workspace that
- 7:51
looks like this. She's really into
- 7:52
fairies and like drawing and mushrooms
- 7:55
and she's like expressing her whole self
- 7:57
and her creativity online in this new
- 7:59
way that we kind of really haven't had
- 8:01
as regular people for a long time since
- 8:03
like the '90s.
- 8:06
And under the hood, like all of those
- 8:09
sites are just the tip of the iceberg.
- 8:11
So, when Anthea gets somebody who's
- 8:12
interested in a retreat, Zo texts her
- 8:15
with their number.
- 8:17
Anthea now immediately calls them at
- 8:19
that moment of intent to buy. She texts
- 8:21
her Zo, "Give me a payment link." And
- 8:23
she closes the deal right then and
- 8:24
there. And that is how she is actually
- 8:26
booking much more revenue than before
- 8:28
with these retreats. Like her retreats
- 8:30
are now like actually kind of hard to
- 8:32
book because there's too much demand.
- 8:35
And that's amazing. And under the hood,
- 8:36
she has a database for all of her
- 8:38
retreats. She doesn't really know what a
- 8:39
database is, but she has one. She has
- 8:41
all of her notes, all of her accounting,
- 8:43
and it's all in one place. If she has
- 8:45
like some images from a retreat and she
- 8:46
wants to put those on her website, she
- 8:48
just tells her Zo to do it and it's like
- 8:50
right there. She can like host files
- 8:52
directly from her personal server. And
- 8:54
she is self-hosting everything.
- 8:58
So, Zo is made to be simple enough for
- 9:00
anyone, for people like Anthea, for my
- 9:01
parents, for other people's parents.
- 9:03
People are always telling me on like
- 9:05
LinkedIn or X that like they sign up
- 9:07
their parent for their parents to Zo and
- 9:09
their parents are like five coding up a
- 9:10
storm. It was awesome to see. I just got
- 9:12
this message on LinkedIn the other day.
- 9:14
Our caterer uses Zo um to kind of manage
- 9:16
his staff at his restaurant. Our
- 9:18
recruiter uses Zo as well. He's really
- 9:21
into it and he's like kind of building a
- 9:22
whole He's like rebuilding his his stack
- 9:24
for his recruiting agency on Zo. Our
- 9:26
marketing agency has become Zo-pilled.
- 9:29
So, it's made for everybody.
- 9:31
But it's also powerful enough for the
- 9:33
audience in this room.
- 9:35
It can be the home for your Open Claw or
- 9:36
your Hermes. You can have a really
- 9:38
nicely set up server to host your
- 9:41
personal agent in if you use one of
- 9:42
those tools.
- 9:43
You can bring your codex or Gemini or
- 9:46
etc. subscriptions and you can build and
- 9:48
host anything. You have like root access
- 9:50
to your own server to this very nicely
- 9:52
set up Linux VM where like the
- 9:54
networking is also like very well set up
- 9:56
and you don't really have to worry about
- 9:57
any of it. You can SSH in and you can
- 9:59
control your Zo via API or MCP. So, very
- 10:03
flexible.
- 10:05
So, scan this QR code if you haven't yet
- 10:07
uh to get $100 in AI credits. Um
- 10:11
just try Zo. I I think you'll like it.
- 10:13
And I'm going to do a little demo and
- 10:15
then I'll have a little bit more to talk
- 10:17
about after that. Um
- 10:19
>> [snorts]
- 10:20
>> So, really quick demo of Zo. You already
- 10:22
saw it in the beginning, but Zo is this
- 10:24
like nice kind of cloud workspace. I
- 10:26
made this website just now. I make all
- 10:27
sorts of websites on the fly.
- 10:29
Um
- 10:30
you can chat with really any model as I
- 10:32
mentioned. You can bring your own API
- 10:33
key. You can even talk to Claude code in
- 10:35
here.
- 10:36
>> [snorts]
- 10:36
>> You've got this nice file system,
- 10:38
built-in cloud storage because it's your
- 10:39
server. You can just like use it as a
- 10:40
Dropbox.
- 10:41
Um and you know, you can obviously like
- 10:43
work agentically in the chat with any of
- 10:45
your files.
- 10:47
You've got automations. So, you can kind
- 10:49
of run tasks on a schedule with AI.
- 10:51
>> [snorts]
- 10:52
>> You've got tons of built-in
- 10:54
integrations. You don't have to set any
- 10:55
of this up.
- 10:56
We have [snorts] lots of skills, a large
- 10:58
skill library. There's a built-in
- 10:59
browser, so you can log in to sites and
- 11:02
let your Zo like buy things for you or
- 11:05
do whatever. Actually, I always text my
- 11:06
Zo these days to buy stuff on Amazon.
- 11:08
It's like really dangerous for impulse
- 11:09
purchasing. I like see something I'm
- 11:10
like, let's buy it.
- 11:12
>> [snorts]
- 11:12
>> Um and as I mentioned, you can host
- 11:14
stuff. So, you can really host anything,
- 11:16
any arbitrary service, HTTP or TCP. And
- 11:20
we also give you a really nice built-in
- 11:21
personal website. This is my Zo space
- 11:24
and this is all set up nicely and it's
- 11:26
got like, you know, nice kind of like
- 11:28
live coding features. Um and you just
- 11:29
chat with it to edit it. And I've built
- 11:31
all sorts of cool things inside of here.
- 11:34
Um this for example is like my Calendly
- 11:37
replacement. People can book time with
- 11:38
me and it has just like this custom
- 11:40
setup that I like. Um and my Zo can like
- 11:42
review the people that are interested. I
- 11:43
don't want anybody like booking time
- 11:45
with me directly.
- 11:46
Um
- 11:47
and so that's kind of like what Zo is
- 11:49
and I want to conclude by just talking a
- 11:51
little bit about the future. So
- 11:54
the reason we're building Zo is to build
- 11:56
a future and better internet, one that
- 11:59
looks more like this where
- 12:01
everybody, [snorts]
- 12:02
every individual has a real home, a
- 12:04
presence on the internet. And [snorts]
- 12:06
companies already have this, but like we
- 12:08
should interact directly with other
- 12:09
people on the internet without
- 12:11
middlemen. And we should also interact
- 12:13
with like the various companies and
- 12:14
their AIs. Um and it should be like more
- 12:17
direct.
- 12:20
Another way to look at this is that like
- 12:22
in the future we will probably just be
- 12:24
interacting mostly with agents and the
- 12:27
agents will kind of interact with the
- 12:28
cloud, like these like software 1.0
- 12:31
things like websites and APIs.
- 12:35
But [snorts] if you dig a little deeper
- 12:36
into here, what's probably actually
- 12:38
going to happen is that most of these
- 12:40
agents that we interact with will also
- 12:41
be in the cloud. Some of them will be
- 12:43
local, but many of them will be in the
- 12:44
cloud. And when those agents are in the
- 12:46
cloud, the question again is like whose
- 12:49
cloud is it?
- 12:51
>> [snorts]
- 12:52
>> So
- 12:53
Claud tag
- 12:54
is newly out and and Claud tag is
- 12:56
actually, if you think about it, a cloud
- 12:59
agent. It's this like cloud Claud. It's
- 13:02
like a company level Claud that has like
- 13:04
all the company's context and [snorts]
- 13:06
anybody can interact with it. And you
- 13:08
know, it's it's an interesting and very
- 13:09
like ergonomic model. Town and Victor
- 13:11
are also kind of similar instances of
- 13:13
this.
- 13:14
>> [snorts]
- 13:15
>> But
- 13:16
the problem with Claud tag is this.
- 13:19
Whose cloud is it? Whose Claud is it?
- 13:21
It's not really your company's Claud.
- 13:23
And as you kind of improve your Claud
- 13:26
tag setup, like what really improves is
- 13:29
Anthropic, right? Like the the arrow is
- 13:32
pointing the wrong way.
- 13:35
And [snorts] this is intelligence
- 13:37
feudalism.
- 13:38
The intelligence now is bubbling up the
- 13:41
wrong way.
- 13:42
And you are still a peasant again, or
- 13:44
your company is still kind of
- 13:47
a peasant in a way.
- 13:49
>> [snorts]
- 13:50
>> And at Zoe believe
- 13:52
really strongly in changing the way that
- 13:55
things work and giving people and
- 13:57
companies more [snorts] ownership over
- 14:00
their cloud and over their intelligence.
- 14:02
So, we believe that the future should
- 14:04
actually look more [snorts] like this.
- 14:07
Anybody, a company or an individual
- 14:10
should be able to publish agents and own
- 14:12
them and have end users and like have
- 14:14
the end usage of their agent improve,
- 14:17
like self-improve their own kind of
- 14:19
published agent. And that is the future
- 14:21
that we're building. We have [snorts] a
- 14:23
kind of beta version of this. So,
- 14:25
um I'm going to go back to to this Zoe
- 14:27
space uh and scan this QR code finally
- 14:30
if you want to talk to me, get AI
- 14:33
credits, sign up for the beta of this
- 14:35
kind of new agent publishing platform.
- 14:39
>> [snorts]
- 14:39
>> And uh yeah, also you can check out all
- 14:41
of the speakers that are presenting on
- 14:43
this Zoe space page. Um I'm Ben from Zoe
- 14:46
computer. Thank you.
- 14:48
>> [applause]