Summary
Gamma co-founder and chief product officer John Nurona joins Greg Isenberg to explain what an agent harness is and why it matters for the next wave of AI startups. Gamma, the presentation, document and website tool, has reportedly reached $100 million in annual recurring revenue and recently relaunched as a creative agent.
Nurona describes Gamma’s shift from hand-stitched chains of single LLM calls to a single agent loop that does work, inspects it, and uses tools to deliver finished output. He explains the harness as the set of tools an agent uses to act in its environment, check its own results, and learn a user’s preferences, and compares it to Claude Code, which serves as a harness for programming. He also argues that the so-called GPT wrapper framing undersells the work, since the harness is most of the product.
The conversation then turns to opportunity: where professional consumers are starting to pay for AI, how to break into B2B spending, and what a vertical harness for a niche could look like. Nurona says he would like a Slack agent that learns his voice by observing his messages and drafts first replies. The episode is sponsored by Anthropic and is useful for founders and builders thinking about agent products.
📺 Source: Greg Isenberg · Published October 28, 2026
🏷️ Format: Podcast
