← All organizations

AI agent development and optimization

Mutagent

Mutagent builds tools for engineers developing and operating AI agents. Its workflow connects agent building, evaluation, diagnosis and optimization: teams can turn operational context into an agent specification, scaffold an agent with evaluations, and derive test datasets from production traces. Mutagent Diagnostics identifies problems such as repeated agent loops, broken tool calls and ambiguous instructions, then proposes fixes for human approval.

Its co-founders include Dorian Schlede and Burak Ozafsar. The team's approach centers on analyzing populations of traces: clustering execution histories to find shared failure causes, locating them in prompts or code, and attaching evidence to proposed remedies. Confirmed failure classes are retained for subsequent runs. The tools run locally within an existing coding agent and work with tracing sources such as Langfuse and OpenTelemetry.

Mutagent offers a free CLI tier for optimizing up to three prompts with a user-selected model. Its 12-week design partnership helps teams co-develop workflow tools, observability connectors, evaluation datasets and feedback loops. This combines installable tooling with hands-on engineering work for production agents. Multi-agent optimization remains in design partnership, and the company states that its agent runtime is not yet available.

mutagent.io

1 talk

Newest first

2 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Messages from the stage

Evaluation beyond the score

The session covers agent specifications, evaluation datasets and metrics, calibrated LLM-as-a-judge scoring, and harness effects as parts of the evaluation workflow.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28