Unique Research · Startups · December 9, 2024
Y Combinator held Demo Day this week for its first Fall batch — 95 startups, heavily AI (by one count ~87%). Like Summer and Winter batches this year, customer-service AI and AI agents loom large. Four to watch:
HumanLayer
What: APIs so AI agents can call humans for help and approval when needed.
Why: Agents boost productivity only if they behave. Human-in-the-loop prevents runaway agents; too much oversight kills the efficiency gains. HumanLayer intervenes only when needed.
Raycaster
What: Research agents for enterprise sales teams.
Why: Rare excitement for B2B lead-gen software. Raycaster digs for specific details — lab equipment a company uses, what the CTO said at a recent conference — to sell at the right time, the right way. Stands out among tools that still aggregate surface info.
Galini
What: Compliance guardrails for AI apps.
Why: Helps enterprises set AI guardrails against company policy and regulation — and lets them own the controls and evaluate whether the rails work.
CTGT
What: Tooling for enterprise customers to manage AI hallucinations.
Why: Hallucinations are hard and unsolved. CTGT won’t catch everything, but active monitoring and review of enterprise models to spot anomalies is a useful upgrade — and Fortune 10 testing suggests buyers want it.
— END —