Introducing Adapt-1 Machina: Learning Continuous Control Sequences
With Adapt-1 Machina, we explore how continuous sequences of machine controls can be learned through reinforcement learning, starting with coarse outcome feedback and without demonstrations or a critic.
Evolutionary methods in RL often ignore which actions each candidate takes during its episode.
Today we explore Reinforcement Learning over Executed Trials: Adapt-1 revises the commands that ran and scores each revision against the run it came from. 🧵
Introducing CLP: coupled learning and planning from limited experience and delayed outcomes. Learning guides trials; planning and execution generate evidence.
We study a response to inevitable training-data scarcity, starting with adversarial decision-making and machine control.
Can Adapt‑1 Preview stay in an interactive decision loop while thousands of outcomes keep changing its persistent decision state?
We finished migrating to the infrastructure intended for public access. Here’s what latency looks like as retained experience accumulates.
8480_Howard retweeted
Excelent research, a must read for the whole crypto space on X.
"A defining trait: valuations have basically nothing to do with revenue. plenty of these crossed $1B while still pre-product. Some pre-thesis"
Shows you how mispriced $REI still is.
WTF is an AI NeoLab?
A neolab is basically a pre-revenue startup chasing some long-term AI breakthrough, usually valued at $1B +. Research first, product later (if ever). As of mid-2026 the space is huge. One estimate pegs AI research labs at ~$1.68T in private value, but it's very top-heavy: 93.6% of that sits with four frontier generalists (OpenAI, Anthropic, xAI, Safe Superintelligence).
Examples of the other 6.4%: Isomorphic Labs (~$3.5B), Xaira (~$2.8B), Periodic Labs (~$1.5B), AMI Labs (~$3.5B), Decart (~$3.1B). By May 2026 there were reportedly 63 of them.
A defining trait: valuations have basically nothing to do with revenue. plenty of these crossed $1B while still pre-product. Some pre-thesis.
Non-LLM is a real big bet in this space. 'Continual learning' (models that keep learning after you deploy them, no full retrain) is one of the things today's LLMs genuinely can't do. so a lab that looks close to cracking it isn't pitching "another model," it's pitching a whole new paradigm. Going non-LLM actually helps too, since the LLM lane is where that 93.6% already lives. investors are out hunting the other ~6%, the "next architecture resets everything" bet they always say they want.
The rough ranges. for a pre-revenue neolab, ~$1B is the floor. That's more or less the definition at seed. What moves you up from there:
- Strong team, sharp thesis, no demo → ~$1B (standard procedure)
- A working API or near-product that actually de-risks the core claim → that's the multiplier. A real, novel capability is rare in this crowd (most are pre-product), so it can push you toward $3–8B+ if things such as the continual-learning result holds up and reads clearly to technical investors.
BUT circle back to the weird part: the API matters less as a revenue signal and more as proof the science is real. and honestly the biggest swing factor isn't the tech at all. it's whether investors buy your team as an "existence proof" that can actually reach the frontier. Same tech, two founding teams, easily a 5x gap.
A reward can tell an RL/online learner that something worked without telling it which combination of internal signals made it work.
Today, we’re unlocking two learning rules in Adapt-1 Preview: Counterfactual Utility Plasticity (CUP) and Temporal Context Projection (TCP).
8480_Howard retweeted
With the latest feature unlocks of Adapt-1 we wanted to further solidify its structure.
Temporal Context Projection (TCP) and Counterfactual Utility Plasticity (CUP) are rules that you can toggle on/off.
Temporal Context Projection allows the substrate to learn which parts of past experience are relevant to the present decision. So, instead of treating every observation the same way, delayed evidence now becomes explicit and revisable context.
Counterfactual Utility Plasticity instead asks which signals become valuable specifically when combined. Recording predictions before outcomes are known makes possible to evaluate each combination by the predictive utility it contributes beyond its components. That way useful structures are reinforced, harmful ones inhibited, and stale ones allowed to decay under distribution shift.
This has nothing to do with episodic memory, instead CUP and TCP change how retained evidence is transformed into decision state and organized into persistent predictive structure.
TCP and CUP combined make rules whose temporal reach and internal composition can both reorganize online as consequences arrive. TCP learns when the past matters, CUP learns what becomes meaningful together.
Docs : docs.reilabs.org/docs/neuroa…
docs.reilabs.org/docs/neuroa…
8480_Howard retweeted
Already can tell how dumb this market is when you get a major publication and potential break through and folks decide to sell.
Yeah i get it, it is highly technical and won't matter for most but what $REI is doing is more innovative and on the forefront of AI than any other web3 ai lab.
Grei base:0x6b2504a03ca4d43d0d73776f6ad46dab2f2a4cfd
Already can tell how dumb this market is when you get a major publication and potential break through and folks decide to sell.
Yeah i get it, it is highly technical and won't matter for most but what $REI is doing is more innovative and on the forefront of AI than any other web3 ai lab.
We revisited DeepMind's DM Alchemy, comparing three approaches: Adapt-1's ontology-based online sequential learning, Pinon et al.'s transformer dynamics with tree search, and AlKhamissi et al.'s modified representation with episodic memory.
Full Blog: reilabs.org/blog/ADAPT1-BEHA…
8480_Howard retweeted
about to do our largest campaign yet in the big apple, it’s not just one thing it’s three different activations $WISHBONE
8480_Howard retweeted
the $WISHBONE truck has been spotted a few times now while in Miami, have you seen it?
Man, people have been busy trenching on Robinhood chain for the last two weeks, but these projects are cooking like crazy.
My short summary of what's cooking for some of the better projects on Base:
$REPPO @reppo: new tokenomics, updated Datanet staking mechanism, and V3 is coming. More details will be shared during the upcoming X Spaces. The team also dropped a crazy article, definitely a must-read: “Reppo Fundamentals and Our Q3 2026 Outlook” with a lot of alpha.
teapaw.org/reppo/status/207611631…
$REI @rei_labs: no posts on X, but I've been reading their TG chat every day. The devs are cooking and staying busy. I feel like something big could come out of nowhere. I've been paying attention to this team for a long time, and that's just their style, they don't overhype anything. Be ready.
$SERV @openservai: released two articles covering their vision and plan for banking. V2 is coming, bringing new tech like Shadow Agents and Multipath. There will be an AMA in Telegram on July 16th at 6 PM CET.
teapaw.org/openservai/status/2064…
teapaw.org/openservai/status/2076…
teapaw.org/openservai/status/2075…
$KTA @KeetaNetwork: Keeta is opening its ecosystem to fintechs, banks, and developers. I found a great article about the new fee model.
teapaw.org/guitarquant/status/207…
teapaw.org/KeetaNetwork/status/20…
$LIQ @liquid_launcher: @_proxystudio (dev) posted a lot of important tokenomics updates about @liquid_launcher. The founder of $VVV @ErikVoorhees also followed him on X, and @jessepollak commented that what they’re building is interesting.
$DEUS @xmaquina: voting on Strategic Allocation in Gecko Robotics.
I like $VIRTUAL too, but if you’ve been trenching on Robinhood, you probably already know everything new about it.
Have a GREIt week.
If you're tired of chasing rotations, consider researching $REI, $REPPO, $SERV, $POD, and $KTA.
Every one of them has meaningful catalysts on the horizon.
Man, people have been busy trenching on Robinhood chain for the last two weeks, but these projects are cooking like crazy.
My short summary of what's cooking for some of the better projects on Base:
$REPPO @reppo: new tokenomics, updated Datanet staking mechanism, and V3 is coming. More details will be shared during the upcoming X Spaces. The team also dropped a crazy article, definitely a must-read: “Reppo Fundamentals and Our Q3 2026 Outlook” with a lot of alpha.
teapaw.org/reppo/status/207611631…
$REI @rei_labs: no posts on X, but I've been reading their TG chat every day. The devs are cooking and staying busy. I feel like something big could come out of nowhere. I've been paying attention to this team for a long time, and that's just their style, they don't overhype anything. Be ready.
$SERV @openservai: released two articles covering their vision and plan for banking. V2 is coming, bringing new tech like Shadow Agents and Multipath. There will be an AMA in Telegram on July 16th at 6 PM CET.
teapaw.org/openservai/status/2064…
teapaw.org/openservai/status/2076…
teapaw.org/openservai/status/2075…
$KTA @KeetaNetwork: Keeta is opening its ecosystem to fintechs, banks, and developers. I found a great article about the new fee model.
teapaw.org/guitarquant/status/207…
teapaw.org/KeetaNetwork/status/20…
$LIQ @liquid_launcher: @_proxystudio (dev) posted a lot of important tokenomics updates about @liquid_launcher. The founder of $VVV @ErikVoorhees also followed him on X, and @jessepollak commented that what they’re building is interesting.
$DEUS @xmaquina: voting on Strategic Allocation in Gecko Robotics.
I like $VIRTUAL too, but if you’ve been trenching on Robinhood, you probably already know everything new about it.
Have a GREIt week.






























