zaterka

Thank you and let me know if you have feedback. I'll check out isomorphic, solutions that integrate non-technical teams are always welcome.

There's that, but also economists are most often wrong then right on their forecasts.

For AI subscription, I found Deepseek API to be surprisingly cost-effective, so I'm using that one for personal projects and it's working very well. For work, I use whatever my employer pays for (mostly Anthropic).

I'm also testing some workflows with a larger planing and reviewer model (Opus 5.5, GLM 5.3) and a smaller self-hosted model for implementation (Qwen3.8-27B), also to very good results.

Ando for Harness, I'm using a custom one built on top of the Deepseek Harness, and sometimes ClaudeCode with some add-ons I also built to make my life easier.

As a heavy LLM user in my daily work, I've been working on optimizing agent memory on complex projects. That resulted in this simple Rust-based tool called Wayfinder: https://github.com/zaterka/wayfinder

I've been trying to improve agent knowledge over a codebase, and existing approaches weren't giving the result I wanted, and I wanted to keep it simple. In the end, I added a semantic layer on top of an LLM Wiki and it saved 30% of token usage in multiple benchmarks I ran. It also adds some nice features to deal with concurring agents updating the shared memory, to ensure all information there is reliable and fresh.

It's published to PyPi, so an "uv tool install wayfinder-memory" already does all the work, and it's compatible with almost all Harnesses.