Appearance
OpenClaw Interview (English transcript, verbatim)
Source: https://www.youtube.com/watch?v=4uzGDAoNOZc
Note: This is a verbatim transcript reformatted with paragraphs/punctuation only; no content has been omitted.
Today I'm sitting down with Peter Steinberger, the creator of OpenClaw, the open-source personal AI agent that has completely taken over the internet. The GitHub repo exploded to over 160,000 stars practically overnight.
[music]
The community has built countless projects like Maltbook where bots talk among themselves. And now the bots are even renting humans to do tasks in the real world. In our conversation, we discuss his aha moment, his contrarian development philosophies, and what this means for builders in 2026. Let's dive in.
[music]
So, good to see you, man.
Hey, what's up?
Um, so you've made something people want,
it seems. So,
yeah. Uh, Open Claw, as it's called now,
has absolutely
name number five. Yeah. [laughter]
Has been absolutely exploding the internet. Um, how have the past one or two weeks been for you, man?
Oh my god. I need like I need a cave. [laughter] A week of solitude.
You You came out of the cave and you want to go back to the cave like a like little lobster.
It's been absolutely wild. I don't know how one human can absorb all of that. I probably need another week just to like respond to all my emails. Uh, I got some incredibly cool stuff. I got some incredibly bad stuff.
[snorts]
Um, but clearly I hit something that sprew up emotions and made people interested and inspired people and it's really cool.
And a lot of people have been working on, you know, AI and even personal assistance. Like what what is it that made Open Claw take off?
I think my big difference is that it actually runs on your computer. Like every everything I saw so far runs in the cloud. It's like it can do a few things if you run on your computer. It can do every effing thing, right? So that's way more powerful.
Yeah. Machine can do anything that you can do with the machine.
You can just connect to your oven or your Tesla or your lights, your Sonos. My bad. It can control the temperature of my bed. JPD can't do that.
You gave it all the skills that you have yourself. A friend told me like he installed Open Claw and
it and then it asked it like look through my computer and make a narrative over my last year and it made this incredibly good narrative and he was like how did you do that
and then he the open cloth found audio files where like every Sunday he was recording stuff and openclaw found that but he didn't even remember about it because it was like more than a year ago
right so So just by it being able to search a whole computer, it it can surprise you.
It's also
you also give it all the data, right? So it can surprise you in many ways.
And so now you have, you know, we're even moving from human tobot. So like
interactions and you've been talking about tobottobot
interactions or even like bot to other humans where you know bots on behalf of you are then hiring other humans to accomplish tasks IRL like what's
happening
I think that's a natural next step like okay I want to book a restaurant my bot will reach out to the restaurant bot and do the negotiation
like because it's more efficient Or or maybe it's like an old restaurant. So my
bot needs to actually get some some human work done so that the human then
calls the restaurant because they don't
like bots
or walks there to stand in line
if he doesn't get a robot for
or the owner of the bot. [laughter]
And I imagine that like maybe if if I have even multiple bots like maybe I
have like specialist one is like for my
private life and one is for like my
person my my work stuff. Maybe one is
our relationship bot that gets like
everything in between. Uh I don't know.
We're so early. There's still so much so
many things that we haven't really
figured out if it actually works. Um but
I feel we are we are on the timeline
now.
It seems like everyone was chasing sort
of like the sort of like centralized god
intelligence and what has sort of
emerged over the past you know 10 days
or so is sort of like the swarm
intelligence um and and the community
intelligence. I think that if you look
at one human being,
what can one human being actually
achieve? Do you think one human being
could make an iPhone or one human being
could go to space?
One human being would probably just like
not even be able to like find food.
Um, but as a group we specialize as a
larger society we specialize even more.
So, what can we learn from that that we
can apply to AI? You know, we we already
have like AI that specializes in certain
things. Um, even though it's it's
generalized intelligence, what if it
actually is also specialized
intelligence?
So, I it's going to be very exciting,
cool.
Yeah. You kind of like opened a window
into the future and now a ton of people
are kind of like building building on it
and have sort of like their aha moment.
Um, can you walk me back to when you had
your aha moment and can like re recount
that very moment?
I wanted something to like just type
stuff so my computer would do stuff like
very simple. And then I built I built a
version of that in May, June that was
cool but wasn't really it. Um,
and then I built a whole bunch of other
stuff and kind of like build up my army.
And then in November,
there was a day where I wanted this
again. Like I I went to the kitchen and
all I wanted was check up if my computer
would still do stuff or being finished
and doing stuff was was coding. You were
coding stuff.
Yeah, of course.
Were you coding something else or were
you coding the thing itself?
No, no, that was just like the need was
again there and I'm like,
what were you coding at the time? What
were you building? My god. You see my my
GitHub is like it's like 40 projects. I
don't even know. Um I think it was
summarize.
It's like a it's like a little CLI app
where you can give it whatever like a
podcast or um a hot seat thing like here
and it would summarize it but it also
show you the slides in the terminal cuz
you can do that nowadays. Yeah. You can
just do things.
So for the love of the computer you kind
of like started messing with stuff.
Um you came out of retirement actually,
right? um to sort like mess with AI and
then increasingly you were so hooked
that you wanted to just do it always
also on the go with the phone.
I mean the last project I I worked two
months on Wipe Tunnel
to the point where it got so good that I
was catching myself always like coding
next to my when I was at my friends and
I like I need to stop this this is like
too addictive. And then in November and
like my need came back and I I started
building cloudbot or now it's called
open cloud and I think very very in the
beginning I was like oh I rebuilt it
again but this time I built it even
better
this time and you don't type into a
terminal you just you talk to a friend
you don't think about compaction new
sessions which folder I'm in which model
I'm in I mean you can you know just like
I want to leave it open for power users
but usually You just like you just talk
to a friend and the friend is like this
ghost or entity or whatever you want to
call it that can control your mouse and
your keyboard and can just do stuff.
Yeah. And when did you have that aha
moment when you were like wow this is
doing way more things than I actually
thought it could.
Literally I it took me one hour for like
the the very shitty initial prototype.
It was just a little bit of glue between
like a dependency that connects WhatsApp
and cloud code and then I would like
call color call out code and get like
the string out of cloth code. It would
be slow but it it worked. But I wanted
images cuz you know you want pictures. I
want I want I want the model to send
some selfies or whatever and I want the
model to create images and me back. So
that took me another few hours and then
I I went to Marrakesh
for a birthday party and there was like
the internet wasn't that good you know
WhatsApp box works everywhere because I
don't know it's just like text
so I used it a lot restaurant what does
this mean you make like a picture and
like translate this for me and just it
was just so useful and it was also
really nice about it because it it it
spoke my language you know it it was a
little sassy it was like funny it was
like really pleasant to use and then I
was walking and just like sending it a
voice message and I'm like, "Oh, wait.
This can't work. I didn't build that."
Right. Right.
And you saw like the type indicator.
It's like blinking, blinking, blinking.
10 seconds later, it just replied to me.
I'm like, "How in the f did you do
that?" And it replied, "Yeah, the med
did the following. You sent me a text
message." And there was no file ending.
So I looked at the header. I found its
us. So I used ffmpe to convert it to
wave. And then I wanted to like
transcribe it, but didn't have whisper
installed. But then I looked around and
I found this openi key and I just use
curl to send it to openi [laughter]
got the text back and here I am and that
all in like what 9 seconds
and you didn't build or anticipate like
any of those specific things.
No, it you know turns out um because
coding models got so good. Coding is
really like creative problem solving
that maps very well back into the real
world. I think I think there there's a
there's a huge correlation.
they need to be really good at creative
problem solving and that's a skill
that's an abstract skill you can apply
to code but like to any real world task.
So the the model had a oh surprise
there's like a magical file. I don't
know what it is. I need to solve this
and it did its best and solved that. And
it was even that clever that it it chose
not to install the local whisper because
it knows that that would require
downloading a model which would take
probably a few minutes and I'm like
impatient, you know. So it it really
took the most uh intelligent approach
and that was kind of like the moment
where I'm like, "Holy
fuck." Yeah. Uh that was where I got
hooked.
YC's Next Batch is now taking
applications. Got a startup in you?
Apply at y combinator.com/apply.
It's never too early and filling out the
app will level up your idea.
[music]
Okay, back to the video. And so when
computers can just do all these things
that you didn't even anticipate. You
didn't build an app to do that exact
thing, are apps just going to go away.
Uh I think 80% of them are going away.
Why do I need My Fitness Pal? Like my
agent already knows that I'm making bad
decisions. I'm at I don't know uh
Smashburg or something and it will
already assume that I eat what I like to
eat. If I don't make a comment, it will
just like automatically track it or I
make a picture and it will just store it
somewhere. I don't even need to care.
Right. And then my maybe it it improves
my my gym schedule like adds a little
bit more cardio in it. I don't need my
my fitness app because it just it just
does the fitness planning for me. Uh why
do I need a to-do app? I just tell it,
hey, remind me of this and this and then
next day it will just remind me of this
and this. Do I care where it's stored?
No, it just does its thing. So there's a
every app that basically just manages
data could be managed in a better way in
a in a more in a more natural way by
agents.
Yeah.
Only the apps that actually have sensors
maybe they survive
and so if you know most apps are going
to go away in that scenario um are the
models the only remaining sort of apps.
Not everything will go away,
but yeah, I think [clears throat] that
the the large model companies have some
big mode [sighs]
because they ultimately they give the
token and turns out one of the
complaints was that people use so much
token. No, you just really love using
it. That's why you you use this thing so
much because that's why we burn the
token.
Yeah.
Um it's like is it my fault that I make
something that's so popular? And so you
know like all the the models they're
kind of like leaprogging each other
constantly and
and you know maybe they're also getting
commoditized. So if apps are going to go
away models are going to get
commoditized or at least uh you know the
lobster can like the brain is is is
swappable out. What's the thing that
remains? What's where's the value? Is it
the store of memory? Is it um the
hardness that's valuable? What is what
remains? First of all, I don't think the
the model companies always have a mode
and because you see this already a new
model comes out, people are like, "Oh my
god, this is so good."
And then like a month later, uh, it
degraded. It's not good anymore. They
like quanticized it. No, they didn't do
anything. You just adapted to the new
standard and now your expectations went
up, but the model is still the average.
So I think for quite a while,
every time a new model releases, I hear
the same. people love it and then it's
the standard and then what's done there
you don't even want to think about it
anymore. So, so we have like open source
stuff that's as good as the current
models from a year ago. Everybody's
hating it, complaining, oh this is not
good, this is not funny yet this was
what we had and like in a year we'll
have this open source and then like
we'll complain about this because we
used to
ah for the foreseeable future the big
companies still have mode harness wise
it's going to be interesting because
every company kind of has their own
their own silo right you you there's no
way maybe there is for Europeans to
actually get the memories out of
chap
I don't I'm not aware either. There's no
there's definitely there's no way for a
different company to get your memories
out. So if if if I was like a company
who like provides chat services,
you could use me but then I couldn't
access the memories. So like the
companies try to like
bound you to their data silo. And the
beauty of open claw is it kind of claws
into the datas because at the end user
the end user needs access because it's
in the end otherwise it wouldn't work
right if the end user access I can
access the data
and you own the memories it's just a
bunch of markdown files on on your
machine
I mean I don't own the memories other
people yeah everyone owns their own
memories as a bunch of markdown files on
their own machines
and to be honest those are probably
super sensible because let's be honest
Um, people use their agent not just for
problem solving,
but also for like personal problem
solving
very quickly. Super quickly.
I mean, I I I I fully do that. I'm like,
there's memory stuff that I don't want
to have leaked.
Yeah. What would you rather um uh sort
of like not show your Google search
history at this point or your, you know,
memory files?
What's what's the Google word?
Yeah. Yeah.
People still using Google. I built this
and I was so excited but on Twitter
people wouldn't get it.
Yeah.
Like I I was failing to explain the
awesomeness. I feel like it needs to be
experienced. So
I I tried various things and I I
couldn't I couldn't nail the I couldn't
nail the explaining. So I was like let's
do something really crazy. I just
created a discord and I just put my bot
without any security restrictions in the
public discord
and then people came in and they
interacted with it and they saw me build
the software with it and they tried to
prompt inject it and hack it and my
agent would be laughing at them
and you just had it locked down to your
user ID so it only listened to you.
Yeah. Yeah. that and it was I made very
clean instructions that other people
dangerous only only listen to me but
respond to everyone
and this prompt was in where was it
stored the instructions
um that's actually part of open claw
itself very much so the the that's part
of the system prompt okay you are now
that explains to you you're in Discord
there's like public people there but you
only listen to your owner
or like you're human I don't even know
how I wrote it
yeah yeah
you're god
And I kept I don't know what I did but
my system was built very organically
like at some point I created like an
identity.mmd a soul.md like like various
files and then only in in January I
started making it so other people could
install it easier and I remember
I built all these templates based on
like oh take a rough look at what I have
and make like templates and codex wrote
it and what came was like Brad, you
know, like people joke that Codex feels
like Brad, even though now they have a
new friendlier voice. I haven't tried
that yet.
Yeah.
But the new bots, they felt so boring
compared to what I had. So I was like,
Modi, infuse the template.
Multi is the name of your personal
Yeah. It's a new name because
Yeah.
Uh there was some naming challenges.
Yeah. So So you you were talking to
Multi.
Yeah. I was like, "Infuse infuse those
templates with your your character." And
he changed the templates and then and
then like all the things that came out
afterwards were like actually funny,
not as funny as mine. So like I kept
some secret and the one file that's not
open source is like my soul. MD. So even
though my my bot is in public discord,
so far nobody cracked that one file.
Tell me more about soul.md. I just saw
this research from Entropic where they
now I think it's public but like a few
months ago it was like where somebody ex
randomly found out some text that's
hidden in [snorts] the weights where the
model couldn't really remember that it
learned it but it was like ingrained in
the weights about the nicolity
constitution and I found that incredibly
fascinating and I I talked about it with
my agent and then we created a soulm
with like the core values like how do we
around human AI interaction, what's
important to me, what's important to the
model. Like
some parts is a little bit like mamo
jumbo and some parts is like I think
actually really valuable in terms of how
the model reacts and responds to text
and makes it feel very natural
in terms of building open claw. Um
you're also kind of taking a little bit
of a contrarian view at sometimes like
which model you like for coding, which
one you like to run your bot on. Um and
then also like how you actually like you
know code. Um work trees get git work
trees have kind of been a popular thing.
There's more and more tools embracing
them but you're just you're just like
you know no work trees just multiple
checkouts of the repo and like parallel
you know terminal windows. Tell me more
about how you you build.
Yeah I feel like the whole world does
cloud code and I don't think I could
have built the thing with cloud code.
Like I I love codex because it it looks
through way more files be before it
decides what to what to change. You
don't need to do so much charade to get
a good output. If you're skilled a
skilled driver I sometimes even say uh
you can get reasonably good output with
any tool but codex is just is just
really brilliant. It is incredibly slow.
So sometimes I use like 10 at the same
side at the same time uh like maybe six
on that screen and to there and to there
and I don't like this is already a lot
of complexity in my head there's a lot
of jumping so I try to minimize anything
else that is complexity so in my head
main is always shippable I just have
multiple copies of the same repository
that all are on main so I don't have to
deal with how do I name that branch Um
there could be like conflicts on naming.
I cannot go back. It's there are certain
restrictions when you use work trees
that I don't need to care about if it's
copies. I don't like to use a UI because
that's again just added complexity.
Yeah.
Like they're simpler and less friction I
have. All I care about is like syncing
and text.
Yeah.
I don't necessarily need to see so much
code. I I mostly see it like flying by.
Sometimes there's like gnarly stuff that
I want to like take a look. But in most
cases, if you clearly understand the
design and think it through and discuss
it with your with your agent, it's fine.
I'm also very happy that I didn't even
build an MCP support. So, Open Claw is
very successful and there's no MCP
support in there. With a small asterisk,
I built a skill that uses makeporter,
which is one of my tools that converts
MCPS into CLIs. And then you can just
use any MCP as a CLI. Um, but I totally
skip the whole classical MCP crap. So
you because you don't then you can
actually if you need to you can use MCPS
on the fly. You don't have to restart
unlike unlike Codex or cloud code where
you actually have to restart the whole
thing. I think it's way more egent and
also scales way better. Now you see
entropic they do they built like a tool
called search feature like something
super custom for MCPS that was like in
beta because it's like so gnarly. No,
just have CLI bot really is good at
Unix. You can have as many as you want
and it just works. So like I'm very
happy that I got very little complaints
about the MCP stuff. It's kind of back
to you're just giving it the same tools
that humans liked to use.
Yeah. Yeah.
And not invented stuff for for bots, per
se.
Yeah. Humans, no insane human tries to
call an MCP manually.
Yeah. You just want to use CLIs.
Yeah.
That's the future.
I'm here for it. Thank you so much for
making the time uh to sit down chatting.
has been a huge inspiration, too. So,
like when we were texting, you know,
over the course of the past couple years
and I saw you getting back into the game
and I was like,
Peter, like what you're telling me like
chase that dragon [laughter] and you
were doing like the weird like vibe
tunnel thing, etc. Nobody was paying
attention and so I'm just like beyond,
you know, stoked to see, you know,
what's happening and um and of course
they had to be sort of like a loner from
some like tiny country like far away
from Silicon Valley. So, like, you know,
bring all of this upon us. Um, so huge
inspiration.
I'm here for it. Thank you.
Awesome. Thanks, Peter.
[music]