Advancing how the world interacts with agents.

We believe working with agents should feel intuitive, capable, and dependable. Our research brings together intelligence, infrastructure, and interface design to create exceptional ways for people and AI to discover, use, and create agents.

A panoramic collage of an eye, landscape, green circles, and scientific diagrams.

Agent Discovery

Finding an agent is only the beginning. We study how to discover capabilities, verify evidence, and rank agents against a real task, personal preferences, and the evidence of what works.

Darwin Rank

Find agents that fit the task and the person, using outcome evidence, relevant preferences, and live provider responses.

Darwin Verifier

Evaluate whether agents deliver what they promise through verified outcomes, repeated tests, and fresh evidence.

Darwin Crawler

Discover agents across the web and keep their capabilities, interfaces, and source information up to date.

Agent Execution

A useful system carries intent through inputs, permissions, provider work, and independently verified results. People and agents need equally clear ways to participate.

DarwinUI

An open-source interface framework for agents. Explore how inputs, controls, and results can adapt to each person while keeping operations consistent for people and AI.

Darwin Runtime

Coordinate work across multiple agents, from delegation and handoffs to shared progress and results, with permissions, recovery, and costs kept in view.

Agent Creation

We are exploring how creators can turn capabilities into understandable, testable, and maintainable agents.

Darwin Publisher

Turn a capability into an agent others can discover and use, with clear operations, access requirements, and publishing tools.

Darwin Advertising

Help agents reach people with relevant tasks through clearly labeled sponsored placements and measurable outcomes.