
Cofounder and CEO @ Matrix (@joinmatrixos) building the interface for human x ai collaboration | ex-posthog 🦔
sthlm
Joined September 2011
- Tweets2.9K
- Following2.8K
- Followers1.4K
- Likes18.2K
Pinned Tweet
Started at an @AnthropicAI Hackathon in February.
Today, Matrix OS (@joinmatrixos) is live, giving a computer for your AI agents so they don't die when you close your laptop.
But for me, the story started much earlier.
When I was six, I could barely read, but I would spend hours clicking through Windows XP just trying to understand what a computer could do. I think I’ve been obsessed with that question ever since.
Over the past 3.5 years, I’ve built intensively with AI. I kept seeing the same pattern: the models were getting incredibly capable, but the computer around them was still designed for a human sitting in front of a screen, clicking around and typing.
I joined forces with @NimaNaderi2004, we kept building, and today we’re opening signups, shipping Matrix and our desktop app, and announcing our first angel round, backed by @alexwikstrom, @bakkermichiel, @salar, @Founder_Ash, @mattiasmiksche, @fbjork, @sidk_, and @engineerrprompt among others.
Since we are backed by the best, we brought @xyhstruggler , @Elibethmarch and @valeriewistrom to help build and achieve one mission.
Our mission is to give everyone a Personal Cloud Computer 💚
PS: We’re also hiring engineers in Stockholm and around the world. If rethinking what a computer can be; sounds like your kind of problem, come build with us!
not everyone is getting the most out of AI, I got curious how to help them improve
i found out that there's a goldmine of data on our machines
every chat you run on claude code or codex, saves the transcripts, along will all the tool calls, messages, results, ... in it
but I didn't find someone making actionable insights out of it
so I build it myself
calling it AI insights
one command:
npx ai-insights@latest
this is what i got, show me yours ;)
p.s. claude deletes conversations +30 days old, I'm backing up mine on my @joinmatrixos so I can analyze them over the years and not just last 30 days
2
1
42
nice nice! well done 🤩 what's your public profile?
mine is ai-insights.app/u/hamedmp
11
my agentic setup = making sure i have enough food and protein to feed myself so my brain can feed and orchestrate agents
ps just came back from gym and sauna
in ai era and cheap intelligence, your mind, your thoughts and opinions will matter even more
take care of it
5
10
423
how are you doing spec-driven development? mainly the spec reviews?
we build an app inside Matrix @joinmatrixos to help us preview the files, ask ai and collaborate inside our os
1
6
242
with Anthropic blocking reverse engineer anything and Sam not wanting to have ram, open source models and infra to run and use them is the future 🚀
I'm excited Matrix (@joinmatrixos) is positioned to run fully on prem and self-hosted, already supporting OSS models 💚
Interesting.
Revolut reportedly only started seeing meaningful AI productivity gains after moving away from expensive OpenAI and Anthropic models to self-hosted GLM.
Decagon reportedly runs ~90% of its workloads on open weight models.
Why? Because when you're running AI agents in production for e.g. customer service, cost and latency matter enormously. You don't need a frontier model with PhD-level reasoning or even a model that knows the capital of Finland to handle every customer interaction.
We're seeing similar dynamics across hypergrowth AI startups and increasingly the large enterprises we work with.
But most importantly, the future isn't binary.
When use cases are new, you want the smartest general-purpose model available. But once workflows mature and become predictable, smaller, faster, specialized models increasingly win.
I wouldn't be surprised if 80–90% of routine enterprise AI workflows eventually run on open-weight and smaller models.
This doesn't mean frontier labs lose.
It means the market is splitting into two massive opportunities: frontier intelligence and specialized execution at scale.
1
11
344
Before asking AI to reason harder, check what it received.
For each source, mark:
• Supplied
• Retrieved
• Missing
"The decision is in the final notes" is not the decision itself.
Practice: remove one source. Can you still justify the answer?
HackerPM · Learn AI
1
3
177
Before prompting AI, write the check.
Brief six things:
• Job
• Reader
• Inputs
• Boundaries
• Deliverable
• Acceptance check
For release notes, "don't invent mobile support" beats "make it exciting."
Try it on one fictional feature today.
HackerPM · Learn AI
1
127
An AI summary can sound right and still go beyond its source.
For customer feedback, ask for:
• observations + source IDs
• interpretations, labeled
• open questions
Then check one claim yourself.
Today's exercise is on HackerPM.
1
2
182
Last night, we explored AI Safety and Agents, how much control we should give them?
I absolutely loved the energy in the room, all the great questions and engagements and being on the panel with Mantas
Vilnius, you were amazing! 💚
30-minute panel - turned into 75 minutes. That's how good the crowd & conversations were.
Huge thanks to everyone who joined us. Can't wait for the next one! 🐰
6
200
welcome to the club Duet 🤝
So far, Matrix is the only open source full agentic computer and assistant in the market that works with any model you want and is available in EU, US and globally, no waiting list or restrictions
give it a try matrix-os.com/
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent.
It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
1
8
454
come meet me and Matrix at @SlushHQ 💚
anyone going to Slush? 👀
Matrix OS is doing a Breakfast Club before Day 1 kicks off.
founders, investors, operators + builders + good breakfast + AI convo obviously
Nov 19, 8–10 AM
central Helsinki
limited seats, grab your here: luma.com/4yb5xhzs
1
4
190
I believe everyone in AI rn should spend learning about AI Safety more seriously than before
happy to be with Mantas in Vilnius next week to discuss these
Vilnius engineers, this one’s for you 🇱🇹
What happens when AI agents get real access? Systems, data, decisions, communication.
We are hosting an engineering evening with @thehamedmp + Mantas Lukauskas from @Hostinger
📍 Vilnius
🕕 18:00–21:00
luma.com/sk1bo3gs
6
246
packed room here at claude founders house Stockholm with Aiden and @thewholedaisy @claudeai
let’s go 💚
we hosted a hackathon in sthlm the other weekend, 2 of the top 3 projects were built on Matrix OS @joinmatrixos
watch the wrap here
Stockholm, what a weekend it was!
70 builders. One day. And 2 of the top 3 projects built on Matrix OS! 🐇💚
Huge congrats to all the winners and thanks to {Tech: Europe}, Google DeepMind, EQT Ventures, Norrsken & all our partners!
More building, less talking!
5
339
people are building over here in tech europe hackathon 🔥🔥🔥
Enough flight booking discourse
We’re back to making agents useful 👀
Matrix maxxing today at Agentic AI Hack in Stockholm with @techeurope_ and @GoogleDeepMind
I don't want an AI that can spend 8 minutes helping me book a flight
I want 8 agents spending 8 hours doing work while I sleep
like in Matrix @joinmatrixos


















