miniblog.

I find that I'm choosing AI tools based on the quality of the harness rather than the model. For example, I'm using Claude Code Web because its model of remote VMs is extremely convenient. I'm using Perplexity because it's tuned really well for web searches.

Agentic programming workflows rather remind me of genetic programming. The agent has a validation step that looks like a fitness function, and both run iterative trials.

Playing with optional type signatures in Python, I realise that the return type is the most important to me. I'd much rather have a function with only a return type instead of a function with only parameter types. It's often quick to add too.