@dan_preiss
Pinned post
We鈥檙e talking a lot about long running/horizon agents now but owning the full job and getting these learning loops right is so important to enabling agents to do work over hours or days, through a complex logic chain with many in between decisions and handoffsView quoted post on X
We backed @soothlabs at seed. Here's why.View quoted post on X
The team.
@subail founded and scaled Meta's Pittsburgh AI office and serves on CMU's Robotics Institute faculty.
@rsalakhu Salakhutdinov is a @SCSatCMU professor, a pioneer in modern generative modeling, Apple's first director of AI research, and a disciple of Geoffrey Hinton. Chuck Hoover directed billion-dollar innovation programs at Meta.
$50M round led by @felicis at a $335M valuation. Angel investment from @ylecun and @JeffDean. Andrew Bosworth is advising.
What they're building: a continuously trained world model for long-horizon forecasting. It takes verifiable propositions (Will Brent crude exceed $90/bbl in March 2027? Will US CPI exceed 4% YoY in 2026?) and returns calibrated probabilities with a structured causal timeline of driving events.
Institutions commit capital in the face of deep uncertainty every day, and the stakes keep rising while the tools haven't moved. Strategy still gets set with gut calls, internal models, and the occasional expert opinion. Specialist platforms exist in finance, insurance, and energy, but each operates within a single domain and is rarely measured against what actually happens.
Foresight should be a system, not a skill.
LLMs aren't the answer here. They're optimized to generate plausible text, not to produce probabilities graded against real events. Sooth is built on a different architecture designed for exactly that job, with a causal chain users can interrogate, revise, and stress-test as conditions change.
Congrats to Yaser, Russ, Chuck, and the team.
$BIRD is selling its brand and footwear assets renaming itself NewBird AI and using a new $50M convertible facility to pivot into AI compute infrastructure by buying GPUs.
Going from selling wool sneakers to chasing AI compute might be one of the wildest pivots this cycle.View quoted post on X
