Keep pulling the thread on Eno Reyes.
Factory plans to publish research showing that its Droid agent has passed the threshold of being "self-improving."
Factory's founding strategy was to focus on building AI agent products suitable for enterprise environments.
Factory's Droid product provides tooling to analyze a company's entire codebase to identify barriers to AI agent success, as well as ROI analytics and enterprise controls.
The Factory platform supports nearly every frontier AI model.
The Droid AI agent is optimized for long-running tasks lasting up to an hour through techniques like compaction, compression, and prompt caching.
The Droid AI agent uses Chrome DevTools for autonomous browser interaction and QA, with a native implementation planned for release soon.
Product teams at OpenAI, Ramp, and Block use the Linear platform to collaborate with AI agents.
The Droid AI agent allows users to change the underlying AI model mid-session, enabling workflows like using an Opus model for planning and a GPT model for execution.
Eno Reyes claims that combining an Opus model for planning and a GPT 5.2 model for execution outperforms using either model alone for coding tasks.
The Droid AI agent is capable of picking up an existing codebase's styling, brand colors, and design system components to maintain visual consistency when building new features.
Factory is the only AI agent provider that offers enterprise-level, hierarchical controls for managing which users and teams have access to specific customizations like skills and MCPs.
An Account Executive at Factory is the third most active user of the Droid agent within the company.