Lemma Raises $2.3M to Detect and Fix Silent AI Agent Failures

Lemma Raises $2.3 Million to Tackle Silent AI Agent Failures

San Francisco-based AI startup Lemma has raised $2.3 million in pre-seed funding to help companies detect and fix failures in AI agents that can go unnoticed in production.

The round included Matrix, Y Combinator, Liquid 2 Ventures, Vermilion Cliffs Ventures, Irregular Expressions, Cervin Ventures, Comma Capital, Position Ventures, and Eight Capital, along with angel investors from OpenAI, xAI, Meta, and DoorDash.

Founded by Jerry Zhang and Cole Gawin, Lemma is focused on a problem the founders encountered while building AI agents: systems can pass testing but still fail when deployed in real-world environments without producing obvious errors or crashes.

Monitoring AI Agents in Production

Founded in 2025, Lemma monitors live AI agent traffic to identify problems such as infinite loops, failed tool calls, and incorrect interpretations of user intent.

The platform reviews more than one million agent traces each day to identify patterns behind failures and determine their root causes.

Lemma also aims to go beyond monitoring by suggesting fixes directly in the codebase, allowing development teams to move from identifying an issue to resolving it more quickly.

Addressing Silent Failures

Traditional software failures often generate visible errors or crashes. AI agents can behave differently, producing an output that appears valid while failing to complete the intended task.

Lemma is targeting these silent failures, where teams may not immediately know why an agent performed incorrectly and customers may simply stop using the product.

The company positions its platform as an all-in-one system for identifying and resolving these problems rather than simply providing dashboards or evaluation scores.

Growing AI Agent Observability Market

Lemma is entering a rapidly expanding market as companies deploy AI agents across customer-facing and internal workflows.

Competitors and adjacent platforms include Dash0, FriskAI, and Tsuga, which are developing different approaches to monitoring, observability, and security for AI agents.

Lemma's focus is specifically on identifying failures in live agent behaviour and helping teams resolve those failures.

Funding to Expand Detection Capabilities

The new funding will support Lemma's product development and expansion of its detection capabilities.

The company is targeting teams from seed to Series B that are already running AI agents at meaningful scale.

As AI agents become more autonomous, the ability to understand not only when an agent fails but why it failed could become an increasingly important part of deploying them reliably.