Plurai Raises $13M to Stress-Test AI Agents Before They Go Live

Plurai has raised $13 million in a seed round. Team8 led the deal, with U&I, Velocity, and Mercer also taking part. Plurai announced the raise on November 11, 2026. A Google DeepMind researcher joined as an angel investor.

Plurai builds simulations that test AI agents before they go into production. It borrows the playbook used to prove self-driving cars, running agents through hard edge cases in a safe copy of a company’s systems. The founders came from the autonomous driving world.

What does Plurai do?

Plurai stress-tests AI agents before they go live. It builds a digital twin of a company’s systems, including the knowledge base and the tools the agent will use in production. Then it runs the agent through that copy, pushing it against difficult edge cases to find where it breaks. The idea is simple. You want to catch an agent’s failures in a simulation, not on a real customer.

The company argues that most agents are not reliable enough to run without close human supervision. Plurai wants to change that by testing them hard before launch, the same way self-driving software is tested in simulation before it hits the road.

The round at a glance

DetailInfo
Amount$13 million
RoundSeed
ValuationNot disclosed
DateNovember 11, 2026
Lead investorTeam8
Other backersU&I, Velocity, Mercer

Who invested in Plurai’s round?

Team8 led the seed round. U&I, Velocity, and Mercer also took part, and Reah Miyara, a researcher at Google DeepMind, invested as an angel. The backing from an active AI researcher is a useful signal, since agent reliability is a deep technical problem. For more on who funds this space, see our AI agent investors tracker.

Why do AI agents need testing before launch?

AI agents need testing before launch because reliability is their biggest weakness. An agent that works most of the time is not good enough for real tasks, since a single wrong action can cost money or break trust. This is the main reason many agents still need a human watching them, and it is a leading cause behind the AI agents that have failed or shut down. Testing in a safe copy first is how you find the failure before a customer does.

How does Plurai test AI agents?

Plurai tests agents by mapping every path they can take and checking each one. It builds a digital twin of the company’s electronic systems, then maps all possible agent workflows into a graph. It uses that graph to run evaluations and find fixes for the edge cases where the agent fails. It works alongside existing agent tools like Claude Code, Codex, Copilot, and Cursor, rather than replacing them. Over time, Plurai plans to use the data it gathers to train small, custom models for each customer. In tests with Nvidia and Nebius, the company says those smaller models beat midtier frontier models at a much lower cost.

Why does this round matter?

This round matters because it targets the hardest problem in the agent economy. The first wave of funding built agents. The next wave is about making them trustworthy. Reliability, testing, and oversight are becoming their own category, next to agent security. A seed round led by Team8, with a DeepMind researcher backing it, shows that investors see testing as a real need, not a nice to have. You can see where this fits in the wider agentic AI landscape.

What is next for Plurai?

Plurai will build out its simulation platform and its plan to train custom models from the data it collects. The founders bring strong credentials to the problem. CEO Ilan Kadar and cofounder Elad Levi both worked at the autonomous driving company Nexar, and Kadar worked with Waymo to simulate and evaluate its self-driving software. The challenge is proving that a simulation can capture the messy, open-ended ways a real agent fails, which is harder for software agents than for cars on a road.

Frequently asked questions

How much did Plurai raise?

Plurai raised $13 million in a seed round led by Team8, announced on November 11, 2026. U&I, Velocity, and Mercer also took part.

What does Plurai do?

Plurai builds simulations that stress-test AI agents before they go live. It creates a digital twin of a company’s systems and runs the agent through hard edge cases to find and fix failures.

Why is testing AI agents important?

Agents are often not reliable enough to run without supervision, and a single wrong action can cost money or trust. Testing in a safe copy first catches failures before they reach real customers.

Who founded Plurai?

Plurai was founded by CEO Ilan Kadar and Elad Levi, who both previously worked at the autonomous driving company Nexar. Kadar also worked with Waymo on self-driving simulation.

How does Plurai test agents?

It builds a digital twin of a company’s systems, maps all possible agent workflows into a graph, and runs evaluations across them to find edge-case failures and fixes. It works alongside tools like Claude Code and Cursor.

Keep exploring

Scroll to Top