AIUC Raises $40 Million to Test AI Agents Before They Act
Artificial Intelligence Underwriting Company (AIUC) has raised a $40 million Series A round led by Ribbit Capital. The financing brings its total funding to $55 million, with First Harmonic participating, according to the disclosure. TechCrunch
The company was co-founded by Rune Kvist, an early Anthropic employee, and Rajiv Dattani, who served as chief operating officer of AI safety research organization METR from 2024 to 2025 and remains a board member there. The Series A follows a $15 million seed round. Seed investors included Nat Friedman through NFDG, Emergence, Terrain, and Anthropic co-founder Ben Mann.
AIUC names Cursor, Lovable, Harvey, and ElevenLabs as customers. It has developed a standard called AIUC-1 for validating AI agents.
The broader context here is straightforward. Testing the base model is no longer enough to approve a system for work use. Once an agent can use software tools, remember information between sessions, hand tasks to smaller subagents, and touch live company data, the risk shifts from bad answers to bad sequences of actions. Handling permissions, using tools in the right order, and staying within limits when instructions are vague become the real operational problems. Checks need to live where software ships and runs, not only in lab benchmarks.
Looking at what this means for teams shipping agents, the word underwriting is telling. It points to ongoing measurement and assurance, rather than a one-time certificate. Platform and security engineers will ask familiar questions from past rollouts of powerful software. Where validation fits in CI/CD, the automated pipeline that builds and releases software. What telemetry, or recorded activity data, is kept for audit. How policy violations are defined, versioned, and rolled back. How an agent's identity is tied to its action logs across third-party tools. A named standard suggests AIUC wants that safety layer to work the same across vendors, which would cut integration costs if buyers agree on it.
In my view, the customer list is the detail to watch. Coding assistants, legal workflow tools, and voice and media generation all put agents close to actions that matter inside companies. Those areas face strict rules on data access, reproducibility, and incident review. If validation becomes a standard procurement checkpoint there, it could push the rest of the industry to follow. I have watched this pattern before with cloud security, when controls moved slowly and then became routine. The payoff, if it works, is agents that companies can deploy with more confidence.
