Autonomous (@autonomous_labs) header
Autonomous (@autonomous_labs) profile picture

The AI Hardware Company.

Joined January 2022
Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)Photo by Autonomous (@autonomous_labs)
The best room in the house isn't in the house. We built it as a backyard office. One owner turned it into their studio: guitars on the wall, a mic on the stand, a song waiting. What would you make in yours?
3
2
9
887
Autonomous retweeted
We've prototyped so much hardware over the years at @autonomous_labs. Every so often, I find an old prototype tucked away in the lab. This one is pure delight: an AI radio we built back in 2023 with @LangChain ReAct & @OpenAI GPT-4. CNC. Knobs. Very calm. Very Rams. Very Ive.
2
2
1
12
566
Most robots come with a job description. Lamp comes with a blank one. autonomous.ai/lamp A customer wrote "receptionist." What job would you give Lamp?
4
3
3
37
2,440
Autonomous retweeted
We're rolling out Harness v2 to all Harness devices next week. In this upgrade, we've rethought how humans work with agents. Using v2 internally, we're getting more done than ever. Can't wait to bring it to your Harness device! autonomous.ai/harness
2
2
1
9
1,458
Should we build one with @AMD Radeon AI PRO R9700 next?
Just finished a 4x RTX PRO 6000 build for a customer. It's a great time to own your compute and run local AI. We've had a lot of requests for an AMD Radeon AI PRO R9700 build. Anyone running local AI on the R9700? How's it been? Should we build one next?
3
6
1,037
Teams are heading back to the office. But there’s nowhere quiet to call, meet, and work with AI agents. So we built Spaceship, a soundproof indoor pod.
4
2
8
904
We were wrong about where privacy matters most. We bet on the US, and last week alone we shipped to 5 countries: the USA, Poland, Norway, Spain and Switzerland. Where should the next one go? autonomous.ai/computer
3
15
1,153
Autonomous retweeted
I'm grateful for AI agents. While we're out enjoying the weekend with our families, our agent teammates at @autonomous_labs keep going. They're designing a new AI hardware product, working together as a team, and reviewing each other's work. Have a good weekend, everyone!
5
1
1
12
1,145
Autonomous retweeted
Having a lot of fun designing this Spaceship Pod this weekend with the team (humans and agents). Work has changed digitally. But the physical world, specifically the office, is the same. Time to reinvent the office.
1
1
7
676
A little robot on every desk and in every home.
Lamp is shipping. Yesterday we finished packing the first batch in our workshop. Each box holds a Lamp about to wake up somewhere new: a home, an office or a classroom in 20+ countries. 9 months of raising a kid. Time to let them go. autonomous.ai/lamp
5
628
What we shipped this week. - 🚀 Introducing FocusPod: a quiet room for your office - Your Lamp starts shipping today - Harness now comes in Burgundy Full logs below. What should we build next week?
6
5
16
1,018
5/ Lamp is shipping its first batch today. If you've been on the fence, the first units are going out now. 9 months of teaching her something new every day, all for this moment. Let us know when you meet her. nitter.cf/dee_hw/status/21147302…
Tell me you work in physical AI without telling me you work in physical AI. Welcome to the Lamp Swarm.
1
7
125
6/ Our agent fixed its own bugs on Computer 2. It ran Qwen3.8-27B (Q8) fully on-prem, auto-fixed 3D voxel bugs and rendered the result live. 41k tokens, $0 spent. nitter.cf/autonomous_labs/status…
41k tokens active. $0.00 spent. Zero cloud pings. Watch our autonomous agent run Qwen3.8-27B (Q8), auto-fix 3D voxel bugs, and render live on-prem with Autonomous Computer 2.
7
114
Your agent on the home server asks a question. You answer from your laptop at a café. Who can read it on the way? Only your machines. Harness lets you run AI coding agents on all your computers from one window, end-to-end encrypted.
2
3
9
732
We spend weeks on VRAM math. Then the box arrives and trips the breaker in 5 minutes. 2 GPUs: up to 1,600W. 240V recommended. 4 GPUs: ~3,000W. 240V required. A standard 120V/15A outlet: ~1,800W peak. Check your panel before you check the model card.
3
7
579