Ben Lang, who runs community at Cursor and writes the next play newsletter, has published a ranking of 35 tech startups he says are hiring faster than anyone else over the past 90 days. The metric is blunt: new headcount measured against the size of the team they started with. Small companies that doubled show up near the top.
By Lang's own count, roughly 85% of the list works on AI or autonomous systems. That number comes from him, not from any independent audit, and "AI" is doing a lot of work in a year when almost every well-funded startup describes itself that way. Still, the clustering is real, and it falls into two buckets worth looking at.
Security, mostly because agents are a mess
Five names on the list build AI security tooling: Jazz, Tenzai, Straiker, Gray Swan, and Native. The logic isn't hard to follow. Companies are shipping autonomous agents into production faster than anyone has figured out how to lock them down.
Straiker is the one with public numbers to check. In a late-February press release, the company said it grew from 15 to 40 employees in under a year and claimed 8x growth in six months. Eight-x of what, exactly, the announcement doesn't say, and "6 months" of revenue for a company that started selling in March 2025 is a short runway to extrapolate from. CEO Ankur Shah framed the founding as "a contrarian bet" that security would gate enterprise agent adoption, which is the kind of origin story that reads better after the customers show up.
The robots
The other cluster is embodied AI. Lang points to Skild AI on the model side, Mecka AI for computer-vision datasets, and Allen Control Systems for control software.
Skild is the heavyweight here and a slightly odd inclusion if the metric is really hiring velocity off a small base. The Pittsburgh company closed a $1.4 billion Series C in January at a $14 billion valuation, per TechCrunch, tripling its mark from seven months earlier. That's a company hiring aggressively, sure, but it's no longer a scrappy seed-stage outfit. Mecka AI raised $60 million led by Framework Ventures in early June, which lands it squarely in the recent-momentum category the list is supposed to capture.
The one everyone will ask about
Lang also includes Cursor itself (his employer, worth flagging), prediction-market platform Polymarket, and Fleet, which builds reinforcement learning environments for training AI agents.
Fleet is the most concrete data point in the bunch. The Information reported in April that Fleet was in talks to raise at least $50 million at roughly a $750 million valuation, more than seven times its sub-$100 million seed mark. Bain Capital Ventures was said to be leading, with Sequoia, Menlo, and SV Angel following on.
The growth story is genuinely steep: annualized revenue reportedly jumped from around $1 million at the end of last year to north of $60 million in recent weeks. That $60 million figure is the standard startup trick of taking the latest quarter and multiplying by four, so treat it as a snapshot of momentum rather than a year of banked revenue. The bigger question is durability. Labs are throwing money at RL environments right now, but that demand could consolidate into two or three suppliers fast, and most of these startups would not survive the cut.
Lang refreshes these lists every few weeks. The next one will tell you whether this cohort kept hiring or whether the RL gold rush already started thinning out.
