

Agents At Work 4: the future of coding with Thorsten Ball
13 snips Feb 27, 2025
Thorsten Ball, a software engineer at Sourcegraph and author of influential coding books, dives into the future of coding with AI agents. He shares insights on how AI, particularly through large language models, is transforming software development. The conversation highlights the role of AI in enhancing coding productivity and user interaction tools. Thorsten emphasizes the need for developers to adapt to new paradigms, reshape code practices, and embraces AI tools to improve documentation and commit messages for a seamless coding experience.
AI Snips
Chapters
Books
Transcript
Episode notes
AI Agents: Definition and Functionality
- An AI agent is a large language model (LLM) with access to tools.
- It can modify things outside its context window, using tools based on input.
Sourcegraph's Evolution with AI
- Sourcegraph started as a code search company, indexing code for searchability.
- They evolved, realizing users want answers, not just search interfaces, leveraging LLMs to provide those answers.
Thorsten's AI Work at Zed
- Thorsten worked on Zed's AI team, building inline completion using LLMs.
- This experience made him realize LLMs are powerful "fuzzy to non-fuzzy adapters."