Updated October 2026 · what is a now page?
Wicklee — pushing toward first paying customers. Recent work: self-serve checkout, the WES leaderboard, runtime config surface, and a dedicated Models tab. Looking for feedback from anyone running Ollama, vLLM, or llama.cpp in production.
bordercheck — shipped the first version of a data-residency test harness: send a canary through your real AI stack, force the fallback, and prove whether the data stayed inside your border. The first piece of a proof layer for sovereign AI.
Writing — building a body of work on the operator layer of sovereign AI: compound AI on hardware you own, routing on silicon state, and proving your stack stayed sovereign. The Layer Below MCP, The Hard Part of Sovereign AI, and Test the Stack, Not the Diagram are live, with two deeper pieces in draft.
ELM work — finishing a fine-tuned Qwen2.5-7B (568 synthetic examples across six business surfaces, LoRA on DGX Spark, evaluated with DeepEval and LLM-as-judge). GGUF release pending final eval.
Running — base-building, no races scheduled, consistent miles. Tracked on Strava.