SAN FRANCISCO: Raindrop, a startup that monitors artificial intelligence agents for failures, said it has raised a Series A funding round led by CRV, bringing its total funding to $50 million.

Existing investors Lightspeed Venture Partners and Y Combinator participated in the round, along with lead researchers from OpenAI, Anthropic and Thinking Machines. The company did not disclose the amount of the Series A round itself.

Raindrop also announced Simulations, a new product currently in research preview. The company said the tool replays real production traffic and existing test cases against proposed changes to an AI agent’s harness, then applies Raindrop’s anomaly detection to the results. Teams can measure performance against their own test cases while detecting unexpected behavior changes before they reach production, the company said.

“AI engineers can now see ‘what their change will change,'” Raindrop said in a statement.

The announcement comes as AI agents — autonomous systems that can perform tasks on behalf of users — are becoming more capable and complex. The research group METR has found that the length of tasks agents can complete on their own doubles roughly every seven months, and a single run can now last days and involve thousands of tool calls.

“Agents now run for hours, call thousands of tools, and handle real money, real health data, and real customers,” said Zubin Koticha, CEO of Raindrop. “When an agent fails, it does the wrong thing convincingly at scale until someone happens to notice.”

Koticha said Raindrop catches failures in production and, with Simulations, before a change ships. He said the funding will allow the company to bring the testing process used by frontier AI labs to every team building agents.

Raindrop said it detects semantic anomalies in production agent traffic, surfacing issues in traces before users complain. When behavior shifts, engineering teams can see what changed, when it started and which users were affected, with hundreds of real examples attached, according to the company.

Simulations mark the first time companies have access to the same training and testing process used by frontier labs, Raindrop said. OpenAI recently published research on deployment simulation, which regenerates responses to de-identified production conversations with a candidate model to predict misbehavior rates before release. Anthropic builds synthetic universes to train and stress-test its agents.

“I’ve spent a decade backing critical infrastructure for developers,” said Reid Christian, general partner at CRV. “Agents are fundamentally different from traditional software. They are highly capable, autonomous, and non-deterministic. Raindrop treats agent failure as a detection problem, the way a security company would, and Simulations close the loop from production to development.”

Bucky Moore, a partner at Lightspeed, said agents are being deployed across everyday products and high-stakes environments such as defense, where “bad behavior” will increasingly become catastrophic. He said Lightspeed backed the team at the seed stage because of its product vision and customer response, and is “proudly doubling down on Raindrop as the leader in this new category.”

Raindrop said its team includes engineers who invented fraud-transformer models at Robinhood and pioneered malicious anomaly detection at Square, senior security engineers from Segment, Semgrep and Socket.dev, engineers who have led teams at Apple, Airtable and DoorDash, and designers from Apple and PlanetScale. The company was founded by Koticha, Ben Hylak and Alexis Gauba.

Bani Singh, an AI engineer at Vercel, said Raindrop gives his team visibility into severe issues. “If we’re having an issue like a build failure or agents stuck in a loop, we see that issue in Slack,” Singh said. “We see context like how many people are experiencing it and chats we can dig into.”

Raindrop said Simulations are available in research preview and teams can request access at https://www.raindrop.ai/. The company said it is hiring across machine learning, infrastructure, sales and marketing.

Raindrop is a monitoring platform for AI agents. It reads production agent trajectories to catch silent failures, including hallucinated answers, tool misuse and behavior changes introduced by a model upgrade. Raindrop says it is trusted by Fortune 100 and Fortune 500 enterprises and by fast-growing AI companies. It is backed by CRV, Lightspeed Venture Partners, Y Combinator, Figma Ventures, Vercel Ventures and other investors.