06 / Now
I’m in Bengaluru, working on the next Agent Maturity Compass release and writing with Giggr’s founding team. Here’s where those projects and the experiments alongside them stand.
I’m working on AMC v1.2. Its native runtime runs the model loop, mediates tool calls and approvals, and records signed sessions. You can inspect that work in source today. The published npm release is v1.1.1.
See what’s releasedI’m working on our next essay. One question I keep returning to with the founding team is who can make a different decision after a company buys a new tool, and what it takes for them to act on it.
Read our essaysI’m working on the Kaprekar manuscript and its Lean proofs. Whether the maths can help flag an unreliable model answer remains an open question. The Indus script study is an unreviewed preprint. URP and RDLLM have published code for testing compute reductions and AI source attribution.
See the code and papersI’m building ADHD Life OS around the moment you need to choose one task with the energy you have. Style Second Opinion is a private prototype for short calls where verified women can ask volunteers for a second pair of eyes. Both projects are pre-release.
Explore the projectsLocal time