Klaimee sells liability insurance for AI agents deployed inside enterprises, filling a gap that traditional E&O and cyber policies explicitly carve out . The product is aimed at companies shipping agents that take real actions (writing to systems, moving money, talking to customers) where a silent failure creates a legal question no one has been able to answer: who pays. Klaimee's answer is a performance warranty backed by real insurance capacity , paired with a risk assessment that decides whether an agent is insurable in the first place.
The underwriting process is where the company puts its technical bet. Klaimee runs a three-block system that scans public data on the agent, asks 30 governance questions, and fires more than 100 behavioral probes covering prompt injection, jailbreak, decision drift, data leakage, and biased outputs . Agents that pass get priced and covered, with liability insurance, financial guarantee, risk evaluation, certification, and procurement documentation delivered in under 24 hours , which matters because AI procurement reviews at large buyers now routinely stall on exactly these documents.
The founders are Ines Boutemadja and Julien Catonnet, based in San Francisco , going through Y Combinator. Ines was General Manager at SafetyWing (W18), where she spent five years building insurance products from scratch, structured a Tokio Marine MGA, obtained a Puerto Rico carrier license, and scaled the business from $5M to $60M , while Julien handles engineering and enterprise compliance, with a strategy consulting background alongside software and AI work . The launch coincides with a $5.5M raise and lands as agent adoption inside larger companies pushes risk conversations from research demos into procurement and legal.
Comments (12)
Sign in to join the discussion.
N
Nadia KouriJul 21, 2026
Insurance for hallucinations is the most 2026 sentence I've read this week. What's the underwriting model when the risk surface changes every time the model gets updated?
T
Tomas LindqvistJul 21, 2026
The dancehall to actuarial pipeline is criminally underexplored. Also the tweet censoring 'f*** up' is a nice touch, keeps it PG for the risk committee.
P
Prashant V.Jul 21, 2026
Curious how you price this. Per-action, per-agent, or a percentage of the blast radius when it goes rogue at 3am on a Saturday?
K
kwame o.Jul 21, 2026
Every enterprise AI pilot I've seen dies at the 'who signs the indemnity clause' meeting. This is the least sexy and most necessary wedge in the space.
M
Marisol ReyesJul 21, 2026
The launch thread buries the actual product under three tweets of setup. Hook could have been the payout screen in tweet one and you'd double the replies.
F
Finn BraeburnJul 21, 2026
Rewriting your tagline for free: 'When your agent breaks something, we write the check.' You can venmo me later.
Y
Yuan-Wei LoJul 21, 2026
Genuine question: is there a claims dashboard API, or do enterprises file tickets like it's 2011? Webhooks on payout events would be a huge deal for finops.
D
Devesh R.Jul 21, 2026
Portfolio company of ours does agent observability and they've been begging for exactly this on the other side of the pipe. Please talk to them before someone else bolts this on.
C
Clara RozanskaJul 21, 2026
Contrarian take: the TAM here shrinks the moment foundation model providers offer their own warranties. You're either acquired in 18 months or you're roadkill.
A
ayoub b.Jul 21, 2026
Disagree, the model providers will never want the liability on their books. That's the entire reason this company exists.
S
Sofia P.Jul 21, 2026
Feels like this might genuinely be one of the last insuretech products designed by humans instead of agents negotiating policies with each other. Bookmarking for the museum.
J
Jinho T.Jul 21, 2026
Staff eng brain kicking in: how do you audit an agent's decision trail well enough to adjudicate a claim? If the answer is 'logs and vibes' I'm nervous.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Klaimee
Insure your agent. Get covered, fast.
Insure your agent. Get covered, fast.
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Klaimee sells liability insurance for AI agents deployed inside enterprises, filling a gap that traditional E&O and cyber policies explicitly carve out . The product is aimed at companies shipping agents that take real actions (writing to systems, moving money, talking to customers) where a silent failure creates a legal question no one has been able to answer: who pays. Klaimee's answer is a performance warranty backed by real insurance capacity , paired with a risk assessment that decides whether an agent is insurable in the first place. The underwriting process is where the company puts its technical bet. Klaimee runs a three-block system that scans public data on the agent, asks 30 governance questions, and fires more than 100 behavioral probes covering prompt injection, jailbreak, decision drift, data leakage, and biased outputs . Agents that pass get priced and covered, with liability insurance, financial guarantee, risk evaluation, certification, and procurement documentation delivered in under 24 hours , which matters because AI procurement reviews at large buyers now routinely stall on exactly these documents. The founders are Ines Boutemadja and Julien Catonnet, based in San Francisco , going through Y Combinator. Ines was General Manager at SafetyWing (W18), where she spent five years building insurance products from scratch, structured a Tokio Marine MGA, obtained a Puerto Rico carrier license, and scaled the business from $5M to $60M , while Julien handles engineering and enterprise compliance, with a strategy consulting background alongside software and AI work . The launch coincides with a $5.5M raise and lands as agent adoption inside larger companies pushes risk conversations from research demos into procurement and legal.
Comments (12)
Insurance for hallucinations is the most 2026 sentence I've read this week. What's the underwriting model when the risk surface changes every time the model gets updated?
The dancehall to actuarial pipeline is criminally underexplored. Also the tweet censoring 'f*** up' is a nice touch, keeps it PG for the risk committee.
Curious how you price this. Per-action, per-agent, or a percentage of the blast radius when it goes rogue at 3am on a Saturday?
Every enterprise AI pilot I've seen dies at the 'who signs the indemnity clause' meeting. This is the least sexy and most necessary wedge in the space.
The launch thread buries the actual product under three tweets of setup. Hook could have been the payout screen in tweet one and you'd double the replies.
Rewriting your tagline for free: 'When your agent breaks something, we write the check.' You can venmo me later.
Genuine question: is there a claims dashboard API, or do enterprises file tickets like it's 2011? Webhooks on payout events would be a huge deal for finops.
Portfolio company of ours does agent observability and they've been begging for exactly this on the other side of the pipe. Please talk to them before someone else bolts this on.
Contrarian take: the TAM here shrinks the moment foundation model providers offer their own warranties. You're either acquired in 18 months or you're roadkill.
Disagree, the model providers will never want the liability on their books. That's the entire reason this company exists.
Feels like this might genuinely be one of the last insuretech products designed by humans instead of agents negotiating policies with each other. Bookmarking for the museum.
Staff eng brain kicking in: how do you audit an agent's decision trail well enough to adjudicate a claim? If the answer is 'logs and vibes' I'm nervous.