Descriptions:
Eno Reyes, co-founder of Factory AI, joins Peter Yang for a live demo and candid conversation about Droid — Factory’s enterprise-grade AI coding agent that competes in an increasingly crowded field alongside Cursor, GitHub Copilot, and other tools. Factory has been building agent infrastructure for two and a half years with a deliberate focus on enterprise readiness: audit controls, ROI analytics, and full codebase analysis to identify what’s blocking agents from succeeding at scale.
The live demo builds a speed-reading web app using Droid in a Ghosty terminal session, showcasing one of Droid’s distinguishing behaviors: an upfront “grounding step” where the agent reads existing layouts, CSS files, and component libraries before writing a single line of code. The result is a new app that automatically uses Factory’s brand colors, typography, and component borders — without any explicit instruction or custom skill invocation. Reyes describes this as critical for enterprise use cases where consistency with existing design systems is non-negotiable.
Reyes also shares that Factory’s philosophy is “measure twice, cut once” — iterative validation before execution — which he argues produces meaningfully better output than agents that generate and fix in rapid loops. He makes a notable claim that Droid has crossed what Factory calls the “self-improving” threshold, and points out that the company’s third-highest Droid user is a sales AE, illustrating how the tool has spread beyond engineering. The conversation also demystifies agent terminology (skills, hooks, sub-agents) for newcomers evaluating whether to adopt a terminal-based agent workflow.
📺 Source: Peter Yang · Published February 15, 2026
🏷️ Format: Interview
