I just incinerated $10 running GLM 4.7 on Cerebras to try and Ralph a clone of this game I used to play. Took about 15 minutes for the balance to go to $0. Result, not awesome. Maybe okay.
Running Claude Code on qwen3-coder:30b via Ollama does not work.
Internal adoption charts have been vertical: ~30% of all pull requests merged to our frontend and backend repos are written by Inspect. It only took a couple months for us to reach this level of...
Experimenting with a personal agent using the file system, SQLite, skills tools via MCP. So many ways to make things work these days.
Tried to make a new Github account today and the CAPTCHA was absurd.
Still enjoying working on Tarn.
Paul Simon on "what makes you stuck?" Everywhere I went lead me where I didn't want to be, so I was stuck.
Agents are increasingly being trusted to do quantities of work that cannot be verified at code-level detail. Finding verification criteria for the products of work done like this will be critical for...
I've been playing around with using Voice Activity Detection to record multiple separate audio channels, then stitching those together into a transcript with Whisper to do speech to text.
I'm working through my second implementation of an idea that has been bouncing around my head for a while. The general idea is to be able to transcribe a conversation in real time and have an agent...