Greg Isenberg treats GitHub as an early signal for products, services and workflows that may become mainstream later. His selection focuses on repositories that a builder can install now, use to save time or improve a product, and potentially turn into a paid service.
The examples include a writing skill that removes recognizable AI phrasing while preserving the author's ideasText style transfer rewrites language to change its tone or stylistic properties while preserving the intended meaning., and an agent-oriented CRM that lets software research contacts, enrich records and keep relationship data useful without relying entirely on manual updates. Isenberg recommends starting with human ideas and using AI as an editor or operatorHuman-AI collaboration combines human judgment and accountability with AI speed, generation, analysis, and tool use in a shared workflow. rather than asking it to invent everything in one shot.
He also stresses operational risk. A skill scanner can inspect files before installationAn agent skill audit examines an AI agent skill for unsafe instructions, risky code, excessive permissions, data exposure, and dependency problems., including a static mode that avoids sending private contents to a modelData privacy governs how personal, confidential, or sensitive information is collected, used, shared, retained, and protected in AI systems., while a phone-control harness lets coding agents test real iPhone and Android flowsMobile app testing verifies that an application behaves correctly, securely, and consistently across devices, operating systems, and real user flows. through existing device tools. The broader opportunity is to validate one useful workflow first, then decide whether it should remain an internal advantage or become a product for others.
Watch on YouTube
