Towards spatial intelligence, founding @antimlabs
San Francisco, CA
- Tweets 76
- Following 359
- Followers 104
- Likes 744
The two bets that will change robot learning:
→ In-context learning from a single demo
→ Pretraining entirely in simulation
No massive real-world datasets required
Today we’re out of stealth, and we’ve raised a $40m seed round co-led by @creandum and @usv, to bring Instant Robot Learning to life.
For the past few months, we‘ve been busy building a new company.
Machenta.
This follows results from a new kind of robot foundation model we’ve built, which I’m so excited to share with you today!
Usually, it takes hours/days of data collection and training to teach a robot.
We’ve built a model that learns tasks instantly: one demonstration, no further training.
A model trained entirely in simulation, without a single pixel of real-world data.
A model where teaching robots is so quick and easy, it feels like physical vibe coding.
A model for robots that learn like humans do, from someone like you.
Welcome to the era of Instant Robot Learning.
Thank you @Kathi_Wilhelm, @rebeccakaden, @hkanji, @jonno_evans and all of our angels for your support!
@Machenta_
Learn more: machenta.com/blog/instant-ro…
See below for what you can do with Instant Robot Learning!
1
2
127
Finally moved to SF! Now who's deploying robots for unpacking boxes? I have a very real-world benchmark waiting for you
1
7
3,377
keerti manney retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
885
2,634
19,829
4,554,759
keerti manney retweeted
The Era of Evals is coming to AI robotics.
Today we’re launching Reality Check: our leaderboard & first public robot manipulation benchmark.
14,400 real-world rollouts. Four models. Multiple tasks & data regimes. Object placements. Confidence intervals. FR3 Duo stations.
1/10
60
71
688
119,400
"A rumour that something is possible made it possible" ...Let that sink in
Acceleration is here and will hit the world like a shockwave.
An interesting twist in the whole story is that a rumour that something is possible is what made it possible - that’s it. A little grain of sand that launched a thousand gpu racks.
6
454
now, more than ever, is it extremely important to get into pure math and physics. you can accelerate more than you could have at any other point in history. there is so much to be known and discovered. we shall always be at the beginning of infinity.
13
48
571
13,828
If the same general agent can understand a car, an arm, a humanoid, etc. through their sensor/action interfaces, you start to decouple intelligence from the hardware platform itself.
“LLM as puppeteer” is probably just the beginning. The bigger shift might be robots converging on an abstraction layer where a general agent can express intent without needing to understand the low-level control details of every embodiment.
Almost like a hardware abstraction layer for physical intelligence.
Recently, there have been a lot of impressive demos of AI agents, like Claude, controlling robots.
I wrote a short blog post with my thoughts on the advent of these "robot-use agents."
web.mit.edu/phillipi/www/wri…
I think it's an important change in the trajectory of robotics!
1
1
10
2,994
I think “can an LLM control a robot?” is slightly the wrong question. With enough compute and enough time between actions, probably yes. What’s more interesting is what has to change in the architecture, training, and inference for these models to operate at robot speed.
Robotics operates at multiple timescales. Deciding “Pick up the pan and place it on the stove.” can happen relatively slowly. Maintaining balance, correcting a slipping grasp, reacting to contact, or adjusting a joint trajectory cannot.
A model that has to autoregressively generate a sequence of tokens before every physical correction is paying an enormous latency tax for capabilities you often don’t need at that layer. Making the LLM 2x or 5x faster helps, but it doesn’t really change the shape of the problem.
GPT-6 Astra scored 95% on a robot control task, up from Fable 5.1's 40%, with 6.2x fewer output tokens at 2.3x lower cost. 🧵
8
1
25
9,689
There’s also a fundamental mismatch between language generation and robot actions. Most robotics foundation models already use a VLM-style backbone for perception and world understanding, then a separate action head for control.
Even if LLMs get much better at robotics, I’d still expect the control layer to stay specialized: let the big model understand and plan, then let an action head optimized for continuous control actually move the robot.
1
2
167
On the inference side, you’d want to remove as much sequential work as possible: fewer tokens, parallel action prediction, smaller distilled models, and things like speculative decoding to let cheap models propose actions that a larger model quickly verifies.
2
176
On the training side, you probably want models learning compressed action representations rather than describing every motor command explicitly: action chunks, continuous latent actions, hierarchical policies, and objectives that teach the model to hand off a high-level intention to a much faster controller. You could also distill expensive long-horizon reasoning into smaller policies that execute common behaviors without re-solving the task every few milliseconds.
3
287
keerti manney retweeted
In-Context Learning Results Hint at a “GPT Moment” for Robotics
For general-purpose robots to be useful and economical, you need to be able to teach them new skills on the fly. And now we can see how this will work, with in-context long horizon video demonstrations shown by companies like Generalist, Skild, Rhoda, Rhoda, and RobbyAnt.
The idea is simple: prompt with a demonstration, mixed video and proprioception information. Language isn't enough -- it doesn't specify the problem well enough. And it's starting to really work.
more in blog post
10
20
155
37,885
WHAT DO YOU MEAN IT CAN SKATE TOO? 😭😭
We built a small biped robot you can teach new tricks to.
Train it in simulation, run it on the real thing. Meet Microduck 🦆
$399, shipping before Christmas.
pollen-robotics.com/microduc…
github.com/pollen-robotics/m…
1
12
1,125