The Israeli founders giving AI agents a test drive

After working on autonomous-driving technology, Ilan Kadar and Elad Levi's Plurai has raised $13 million to help businesses make AI agents more reliable

Plurai founders Dr Elad Levi and Dr Ilan Kadar. Credit: Guy Hamoy

Two Israeli researchers who helped develop AI systems for autonomous driving companies including Waymo are applying lessons from the road to another challenge – making AI agents reliable enough for businesses to trust.

Dr Ilan Kadar and Dr Elad Levi are the founders of Plurai, which uses simulations to find and fix problems in AI agents before they reach customers.

The company recently raised $13 million in seed funding and opened their simulation platform to developers. The round was led by Israeli venture group Team8, with investors including CyberArk founder Udi Mokady.

The company has also worked with NVIDIA and AI infrastructure provider Nebius on models for a global enterprise customer. Plurai says the specially trained models outperformed models from OpenAI and Anthropic on the customer’s tasks, at 40 and 44 times lower cost respectively.

Kadar, Plurai’s chief executive who has a PhD in computer science and machine learning, said: “AI systems fail when they hit challenges down the road which no one predicted.

“Plurai brings that same approach to AI agents: continuously challenging them with the hardest situations and rarest events before they reach users, learning from failures, and improving before those failures happen in the real world.”

AI agents are software systems that can carry out tasks, such as handling customer enquiries or making bookings, using information and tools supplied by a business. Their growing capabilities have encouraged companies to experiment, but mistakes, unpredictable behaviour and running costs can make wider deployment difficult.

Chief technology officer Levi, who has a PhD in mathematics, encountered similar challenges while working at Nexar. There, he and Kadar helped develop systems for autonomous driving companies, where simulations allow developers to test unusual and potentially dangerous situations before vehicles encounter them on the road.

The Plurai founders worked at Nexar, helping to build scalable and reliable systems for companies like Waymo 
Plurai builds a digital replica of an agent’s working environment, using a customer’s information, tools and policies. It then runs thousands of simulated users and requests through the system to expose problems, from invented answers and incorrect actions to breaches of company policy and sensitive data leaks.

The results help developers fix failures and train smaller language models tailored to particular tasks. These can run agents at lower cost or check their actions as they work. Problems discovered once an agent is live feed back into further simulations.

Plurai says work with early partners reduced failures reaching users by 82 per cent and running costs by 97 per cent. It also reports cutting development cycles from months to hours.

“AI is probabilistic. Enterprises are not,” said Udi Mokady, an angel investor in the company.

“The founders have spent more than a decade building AI for mission-critical environments like autonomous driving, where systems don’t get second chances. That experience shows in how they approach agents, and it’s what separates AI that can be deployed from AI that can be trusted.”

Previously available to selected customers and partners, the platform now has a free tier. Developers can access it through tools including Claude Code, Codex and GitHub Copilot, while businesses can also test agents built on platforms such as Salesforce Agentforce.

Other investors in the seed round include U&I, Velocity and Mercer. Dr Eyal Carmi, chief technology officer of Microsoft 365 Copilot, has joined Plurai’s advisory board. He said: “For enterprises, evaluating AI agents on real customer data is often blocked by privacy, security and compliance constraints, which makes simulation mission-critical.”

plurai.ai

read more: