Plurai raises $13 million to make AI agents reliable

Plurai's founders, Ilan Kadar and Elad Levi, came from Nexar, an autonomous driving technology company, where Kadar worked with Waymo to simulate and evaluate its end-to-end deep learning driving software. That experience shaped their view that AI agents face a similar reliability gap: easy to demo, hard to make trustworthy and cost-effective.

The startup's pitch rests on a two-part method. First, it builds a digital twin of a customer's systems and maps agent flows to surface edge cases. Second, it uses the data gathered to post-train small, open-weight models tailored to that customer. Kadar argues this can deliver quality comparable to frontier models at an order or two orders of magnitude lower cost.

Reah Miyara, a Google DeepMind researcher who backed Plurai as an angel investor, said the company's edge comes from understanding failure modes in production and generating synthetic data to tune models. As businesses try to optimize AI expense, that approach could prove a real advantage, since IT leaders are seeing that frontier-capable models are out of distribution with their specific data and that the cost is not sustainable long term, Miyara said.