Summary
Theo, creator of T3.gg, shares how a hand injury that made typing painful forced a complete overhaul of his development workflow — and how the adaptations he made have proven useful even for teammates with full use of both hands. The video blends a candid personal account with a practical guide to staying productive on a computer using AI agents and voice input rather than a keyboard, covering tools including Whisper Flow for dictation, Claude Code, and T3 Code.
The core workflow shift Theo describes is learning to give AI agents multi-step instructions far enough in advance that tasks complete without requiring follow-up. He walks through examples like instructing an agent to move a downloaded file to a remote machine before the download finishes, then walking away and getting a notification when the job is done. He argues this kind of ambient delegation is available to any developer but rarely adopted because it requires trusting agents with more autonomy than most engineers are comfortable with.
The second half covers T3 Code’s upcoming orchestrator v2, which Theo describes as a significant architectural overhaul enabling agents to spin up sub-threads, delegate to other models, and run parallel Claude Code instances. Because T3 Code is open source, Theo was able to pull the orchestrator v2 branch and build a local version for testing — a practical example of how open-source AI tooling creates a faster feedback loop between developers and the products they build with. He notes he averages around ten hours a day in T3 Code currently.
📺 Source: Theo – t3․gg · Published September 16, 2026
🏷️ Format: Workflow Case Study
