LLMs from the command line: A mini-hackathon!

Update: It looks like the creator of `llm`, Simon Willison, may join us for a demo via Google Meet! If you'd like to hop on remotely to participate or listen in on demos, head to https://meet.google.com/gnv-zkne-djc at 1 PM EST!

---

One of my daily drivers is the `llm` command-line tool, created by Simon Willison. With `llm`, you can chat with AI models directly from the command line—and so much more.

`llm` supports powerful AI features like embedding, structured outputs, and tool use. Plus, you can chain `llm` calls with other command line tools.

But don't be intimidated! `llm` is easy to set up and use. At this event, I'll give everybody a half-hour crash course to get them set up with `llm`. Then we'll spend an hour or so hacking to see what you can do with this tool.

At the end of the hour, we'll share our coolest results and vote to pick the best. I'll have a $25 Starbucks gift card for the winner.

Some ideas: embed some personal files for semantic search. Add captions to your photos with a vision model. Turn your notes into a dataset for fine-tuning. Spin up a self-hosted DeepSeek API and chat with it from your terminal. Let's see what we can do in just an hour!

Here's a quick writeup of the `llm` features I will demo at the beginning of our session: https://gist.github.com/chriscarrollsmith/4670b8466e19e77723327cb555f638e6