Plurai, an Israeli startup that simulates and stress-tests AI agents before they reach production, said on Nov. 11 that it raised $13 million in a seed round led by Team8. The company announced the funding through PR Newswire alongside the launch of its agent simulation platform.
Plurai says it borrows testing methods from self-driving cars, a comparison TechCrunch also used in its coverage of the round. The idea is to put an agent through simulated scenarios, including the strange ones, before a real customer ever meets it.
Why the pitch lands now
Reliability gaps remain a concern as enterprises deploy agents. A VentureBeat survey of 106 qualified respondents, published in October 2026, found that 42% of those who test AI before release reported failures slipping past that testing.
Plurai CEO and co-founder Ilan Kadar framed the problem in his launch post. "We learned that AI doesn't fail on the happy path. It fails in the long tail," he wrote.
The investor view

Aviad Harell, managing partner at Team8, set out the firm's thesis in its investment rationale post. "We believe simulation will become as fundamental to the agent stack as CI/CD is to software," he wrote. CI/CD, short for continuous integration and continuous delivery, is the automated testing and release machinery most software teams now take for granted.
Udi Mokady, the CyberArk founder and a Plurai angel investor, was quoted in the company's funding release. "AI is probabilistic. Enterprises are not," he said.
A small team, with caveats
CTech reported that Plurai has about 12 employees. The company said it was named a Representative Vendor in Gartner's 2026 Market Guide for AI Evaluation and Observability Platforms. That claim comes from the funding release and has not been independently verified.
Beyond Team8 as lead and Mokady as an angel investor, the funding release did not disclose Plurai's valuation or total funding to date.
A crowded, well-funded category

Plurai is entering a field that is filling quickly. Dealroom reported on Nov. 12 that AI-agent evaluation and observability companies raised $272 million through September 2026, up from $166 million a year earlier. The figure is Dealroom's own category tally, and the outlet did not publish the company-level data behind it.
The competition is established. LangChain's LangSmith offers offline and online agent evaluations, trace analysis, human feedback and prompt optimization. In August, LangChain announced LangSmith Tuned Evaluators, which use a specialized model for production evaluation. LangChain claimed frontier-level accuracy with costs cut by up to 82%, an approach that overlaps with Plurai's smaller, specialized evaluators.
Braintrust, another player in the space, announced an $80 million Series B on Feb. 17, 2026, led by ICONIQ, to expand its production-AI observability infrastructure.
What's next
CTech reported that Plurai plans to hire and expand in the U.S. The company has published open research and code, including its IntellAgent repository on GitHub, and said the new funding will go toward building out the simulation platform it opened this week.
Whether a dozen-person company can hold its ground against better-capitalized rivals is an open question. Its bet, in effect, is that testing agents before launch will matter more than watching them afterward.
