Descriptions:
Felix Rieseberg is the engineering lead at Anthropic responsible for Claude Co-work, Claude Code, the Claude desktop applications for Mac and Windows, and Claude for Chrome. In this conversation with How I AI host Cara, he shares how he actually uses the products he builds and what Anthropic is thinking about as Claude’s interface continues to evolve — an unusually direct window into product intent from the engineer shipping these features.
Rieseberg explains live artifacts in concrete terms: interactive, data-aware documents — pitch decks, dashboards, reports — that can pull in updated information and regenerate without being rebuilt from scratch. He illustrates the concept with a founder use case: maintaining a pitch deck that stays current with the latest user metrics across dozens of individual VC conversations. He also walks through personal projects he has built with Claude, including a furniture planning tool and a physical hardware “Claude Buddy” — a custom Wi-Fi and Bluetooth device with a large approval button that he wanted to press every time Claude needed human confirmation. Claude built the entire firmware and interface in a single shot with no corrections required.
The interview also explores the broader question of why AI adoption lags behind capability. Rieseberg argues the gap is not what models can do but whether users recognize that almost any problem is worth handing to these tools — a mental model shaped by years of software that simply did not work. His 9-year-old’s daily Claude use for cybersecurity exploration, running terminal commands alongside Claude, illustrates what becomes possible for people who approach the tools without pre-learned limitations. A valuable reference for understanding Claude’s product direction from someone directly responsible for shipping it.
📺 Source: How I AI · Published May 25, 2026
🏷️ Format: Interview
