How the engineer behind Claude Cowork actually uses Claude | Felix Rieseberg (Anthropic) · How I AI summary

Key takeaways

AI should handle background tasks to free up human creative energy, rather than just automating simple mouse movements.
Many problems can be simplified and tackled by abstracting them up to AI—such as managing personal inventories from emails or designing house layouts.
Choosing the right AI model depends on how well you can scope the problem; Sonnet 4.6 often suffices for most tasks, while Opus is reserved for more complex issues.
Live artifacts with auto-refresh capabilities are powerful tools for creating real-time dashboards that continuously update from connected data sources.
Design latency thoughtfully, making asynchronous and background processes seamless and delightful for users.
Building simple hardware devices connected to AI (like a $20 'claw') opens up fun, physical ways to interact with AI decisions, making tech more accessible and personal.
Education on how broadly problems can be approached with AI is essential; encouraging users to think creatively expands AI’s impact.
Politeness and framing in interactions with AI can help foster a better mental model and trust, which in turn improves usability.

Summary

Innovative Uses of Claude for Personal and Professional Projects

Felix Rieseberg demonstrates the versatile ways he employs Claude to streamline various aspects of life and work. He discusses designing detailed, interactive virtual home walkthroughs from floor plans, leveraging AI to analyze and create 3D models, and automating personal inventory management by querying emails about purchases like furniture and clothing. These examples highlight how AI can elevate everyday tasks from mundane to highly creative endeavors, emphasizing the benefit of abstracting problems up to AI for simplified solutions and continual system improvements.

Design Philosophy and User Experience with Claude

The episode explores Felix's approach to product design, emphasizing the importance of understanding user intentions at different levels—whether users need quick answers, in-depth insights, or engineering solutions. He advocates for a layered abstraction where users are guided to ask simpler, broader questions that AI can handle, reducing manual data entry and fostering creative problem solving. The discussion touches on the importance of trust, patience, and asynchronous interfaces to make AI interactions more delightful and efficient, including handling latency gracefully.

Technical Insights and Creative Hardware Projects

Felix delves into technical decision-making, such as choosing between different AI models based on problem scope. He also shares fascinating projects like building a $20 IoT device to physically approve AI actions, showcasing how accessible hardware hacking has become. These projects demonstrate democratization in AI-powered hardware, encouraging listeners to experiment and personalize their AI integrations without extensive technical backgrounds, fostering a DIY spirit around innovative AI applications.

Educational Gaps and Future Potential of AI in Everyday Life

The discussion emphasizes the need for better education and awareness about AI's capabilities, urging users to see AI as a tool for automating and elevating a broad range of problems. Felix highlights the potential for AI to alter workflows fundamentally by encouraging people to 'think abstractly' about their problems and structure data and systems more effectively—whether it's managing promises, building dashboards, or creating personalized tools for children. He advocates for changing human-AI interfaces to unlock AI’s full potential.

Show notes

Felix Rieseberg is the engineering lead for Claude Cowork and Claude Code Desktop at Anthropic. He previously spent five years at Slack building developer tools. In this episode, Felix demonstrates how he uses Claude to solve real-life problems: analyzing floor plans to build interactive 3D house walkthroughs, automatically tracking promises he makes on Twitter, and building a $20 hardware device that physically approves Claude actions with a button press.

What you’ll learn:

How to use Claude Cowork to turn a 2D floor plan into an interactive 3D walkthrough where you can move furniture around
The “go one abstraction layer up” philosophy: why you should never manually enter data Claude can find itself
How to use your email as an inventory database for furniture, clothing, and personal purchases
When to use Opus vs. Sonnet 4.6 (hint: it’s about how well you can scope the problem, not technical complexity)
How live artifacts work and why they’re powerful for dashboards that refresh with real-time data from your connectors
The product philosophy behind making latency delightful
How to build your own $20 hardware device using Claude Code (no hardware experience required)
Why Felix never reads the code Claude writes and judges it purely on output

—

Brought to you by:

Magic Patterns—Prototypes that look like your product

Guru—The AI layer of truth

—

In this episode, we cover:

(00:00) Introduction to Felix Rieseberg

(02:40) Felix’s role at Anthropic

(03:25) The multiple tabs in Claude and why they exist

(05:55) Using Claude Cowork to design a new house using floor plans

(09:52) When to use Opus versus Sonnet 4.6

(12:37) Building an interactive 3D furniture planner