Plurai, a startup developing simulation technology for reliable and self-improving AI agents, has raised $13 million in seed funding and announced the general availability of its platform.
The funding round was led by Team8, with participation from U&I, Velocity, Mercer, and more than 20 angel investors and industry leaders, including CyberArk founder Udi Mokady. Dr. Eyal Carmi, CTO of Microsoft 365 Copilot, has also joined the company’s advisory board.
Plurai is building technology that helps businesses identify and fix AI agent failures before they reach customers. Its platform uses simulation and small language models (SLMs) to make AI agents more accurate, reliable, and cost-efficient.
According to the company, its technology has reduced agent development cycles from months to hours, cut failures reaching real users by 82%, and lowered the cost of running reliable AI agents by 97% for its design partners.
In joint work with NVIDIA and Nebius, Plurai also reported that its customized models outperformed leading models from OpenAI and Anthropic in accuracy for a global enterprise customer, at 40 and 44 times lower costs, respectively.
How Plurai’s AI Platform Works
AI agents are becoming easier to develop, but businesses still face challenges when deploying them in real-world environments. Agents can make incorrect decisions, produce inaccurate information, violate company policies, or fail to complete tasks.
Plurai aims to address these problems through a simulation-based platform that creates a digital version of an AI agent and its working environment.
The platform uses a company’s knowledge base, tools, and policies to simulate thousands of realistic and challenging user interactions. This helps identify potential problems, including incorrect actions, hallucinations, security risks, and data leaks before an agent reaches customers.
After identifying failures, Plurai helps generate fixes and trains smaller AI models specifically for each agent’s tasks.
These small language models can handle agent operations at lower costs while continuously monitoring performance and checking whether agents follow company policies.
The platform also uses feedback from real-world interactions to improve agents over time.
Platform Now Available to Developers
Previously available to selected partners and customers, Plurai’s platform is now open to developers and businesses, with a free tier.
Developers can use the platform alongside tools such as Claude Code, Codex, and GitHub Copilot. It also supports enterprise AI agent platforms, including Salesforce Agentforce.
Founders and Industry Experience
Plurai was founded by Dr. Ilan Kadar, CEO, and Dr. Elad Levi, CTO. Both founders have PhDs and experience developing reliable AI systems for mission-critical environments.
Before founding Plurai, Kadar served as VP of AI at Nexar, while Levi worked as an AI architect. They helped develop scalable AI systems for autonomous driving companies, including technology that relies on simulation to test self-driving systems before deployment.
The company’s research also includes BARRED, a method for generating synthetic training data for customized small language models, which was accepted at ICML 2026, a leading machine learning research conference.
Plurai is already working with enterprises across AI-focused and regulated industries, including Fortune 500 companies and major financial institutions.
The company was also named a Representative Vendor in Gartner’s 2026 Market Guide for AI Evaluation and Observability Platforms and featured in its report on emerging practices for testing AI agents.
With the new funding, Plurai aims to expand its platform and help businesses deploy AI agents that become more reliable, accurate, and affordable over time.
