Certification plus insurance is the oldest cyber pattern that still raises Series A money. AIUC is running it on agents.
On September 15 San Francisco’s Artificial Intelligence Underwriting Company (AIUC) announced a $40 million Series A led by Ribbit Capital with participation from First Harmonic, according to TechCrunch and FinSMEs. Prior seed: $15 million from Nat Friedman’s NFDG plus Emergence, Terrain, and Anthropic co-founder Ben Mann among others. Total funding: $55 million (TechCrunch). Valuation: Undisclosed. That blank is Undisclosed.
Co-founders: Rune Kvist, an early Anthropic hire, and Rajiv Dattani, former METR COO from 2024 to 2025 who remains on METR’s board (TechCrunch). Named customers via the company to TechCrunch: Cursor, Lovable, Harvey, and ElevenLabs. That is a buyer-list claim from the startup, not a disclosed ACV table.
Product shape, per FinSMEs and TechCrunch: AIUC-1 — an industry-driven certification standard, evaluation framework, and insurance infrastructure for enterprise AI agents. TechCrunch frames SOC 2 as the muse. A consortium of about 250 security and risk leaders shapes the tests. Agents run through roughly 5,000 tests covering jailbreaks, hallucinations, and data leaks; FinSMEs also names prompt injection and unsafe tool execution, with adversarial simulations and quarterly recertifications. Output: a roughly 100-page report. AI runs the tests; humans verify the final audit (Kvist via TechCrunch).
Thesis quote that matters for underwriters, from Kvist to TechCrunch: banks, hospitals, governments, and militaries no longer decline AI because a model is not smart enough — they decline because they have made commitments about what a system will and will not do, and nobody can currently guarantee that. That is a control-assurance pitch, not a model-benchmark pitch. Dattani’s METR pedigree (performance evals for frontier labs) is the credibility bridge buyers will notice.
POV: SOC 2 became the buyer checklist because auditors, insurers, and procurement agreed on the ritual. AIUC-1 becomes that checklist only if the 250-person consortium keeps shaping tests that CISOs actually put in RFPs — and if the insurance layer pays when an agent fails a certified claim. It is theater if the sticker ships without claim history, without public failure modes, and without a buyer who can refuse a vendor that skipped the audit. Named customers help. Undisclosed valuation keeps the raise honest. Question whether certification-plus-insurance is the durable path to agent adoption or a paid badge on top of agents that still jailbreak in production. Sources: TechCrunch Sep 15 and FinSMEs Sep 15. Valuation stays Undisclosed.
Underwrite sheet — sourced only: $40M Series A led by Ribbit + First Harmonic (TC + FinSMEs); $15M seed NFDG/Emergence/Terrain/Ben Mann among others; total $55M (TC); SF; co-founders Kvist (early Anthropic) + Dattani (METR COO 2024–2025, still board) (TC); customers Cursor, Lovable, Harvey, ElevenLabs (company via TC); AIUC-1 standard + third-party audit/cert + insurance (FinSMEs/TC); ~250 consortium; ~5,000 tests (jailbreaks/hallucinations/data leaks + FinSMEs: prompt injection/unsafe tool execution); ~100-page report; AI tests / humans verify; quarterly recerts (FinSMEs); valuation Undisclosed. The pattern is classic cyber. Valuation: Undisclosed.