Plurai
Category: AI Developer Tools
Plurai builds a simulation and post-training platform that tests AI agents with thousands of simulated realistic and adversarial users, fixes the failures it finds, and trains small language models for each agent. Plurai was founded in 2025. The company is led by Ilan Kadar. Based in Israel. Team size: 11-50. Total funding raised: $13M. Latest round: Seed. Key investors include Team8 (lead), U&I Ventures, Velocity, Mercer, angels including Udi Mokady (CyberArk founder) and Reah Miyara (Google Cloud).
- Founded
- 2025
- Headquarters
- Israel
- Team size
- 11-50
- Total funding
- $13M
Value proposition
Find and fix AI agent failures before they reach customers, then keep agents reliable and cheaper to run in production. Plurai builds a digital twin of the agent and its environment, simulates users against it, generates verified fixes, and post-trains task-specific small language models (SLMs) that act as always-on reviewers.
Products and solutions
Plurai Simulation (digital-twin agent testing with fixes verified against a regression set), Plurai Post-training (custom SLMs for evaluation, guardrails, and agent tasks), plugins for Claude Code, GitHub Copilot CLI and OpenAI Codex (Cursor listed as coming soon), support for agents on Salesforce Agentforce, open-source projects and research papers.
Unique value
Applies autonomous-driving-style simulation to AI agents, so testing, fixing and post-training run as one self-improving loop.
Target customer
Enterprise AI teams and agent developers, including AI-native companies and regulated industries such as financial services (Fortune 500 and major financial companies), plus individual developers via a free tier.
Industries served
Enterprise AI agents, financial services, IT managed services, and other regulated industries
Technology advantage
Original research including BARRED (synthetic training data for custom SLMs, accepted at ICML 2026), SCOPE (compiling policies into stable programs), and serving hundreds of tailored SLMs on a single GPU. Company-reported results: 82% fewer failures reaching users, 97% lower cost to run reliable agents, and a custom model 40x and 44x cheaper than OpenAI and Anthropic frontier models at higher accuracy.
How they differentiate
Competitors focus on evals and observability with test cases. Plurai simulates a full digital-twin world, generates verified code fixes, and post-trains task-specific SLMs for low-cost production enforcement.
Main competitors
Maxim AI, Braintrust, Galileo
Key partnerships
NVIDIA and Nebius (joint post-training work with the NVIDIA Nemotron model family), Eon (benchmarks for AI agents), Salesforce Agentforce support, Gartner named it a Representative Vendor in its 2026 Market Guide for AI Evaluation and Observability Platforms.
Major milestones
Founded 2025, emerged from stealth with a $13M seed round led by Team8 and general availability of its platform on 2026-11-11, BARRED accepted at ICML 2026, named a Gartner Representative Vendor in the 2026 Market Guide for AI Evaluation and Observability Platforms.
Growth metrics
About 12 employees in Israel and the US, planning to double headcount and expand US operations by end of 2026. Platform opened to all developers with a free tier on 2026-11-11.
Market positioning
Positions itself as the reliability and simulation layer of the agent stack, expecting simulation to become as basic to agent development as CI/CD is to software.
Geographic focus
United States and Israel
About Ilan Kadar
Ex-Nexar VP of AI & Automotive and Director of Artificial Intelligence; PhD in computer science / machine learning. Built AI systems for autonomous-driving companies such as Waymo.
Latest news about Plurai
More AI Developer Tools companies
Official website: https://plurai.ai