The Age of Jazz
LLMs and the future of collaboration
Writing
Occasional long-form pieces. No schedule — just when there's something worth saying.
-
-
The Dream is Sci-Fi. The Tech is Catching Up.
Building Mike — a local AI agent with tiered memory architecture. On why personality without memory is theater, and what it takes to build something that actually compounds.
Citizen Scientist
Independent research, AI as co-investigator — because the barrier to serious inquiry just collapsed.
-
LLMs Are Probabilistic. Test Them Like It.
How Karpathy's autoresearch loop generalizes beyond code — and why your AI tool isn't ready after one successful run.
-
Behavioral Variance in Multi-Agent Building Energy Modeling Pipelines
How variance compounds when LLM agents are chained in a processing pipeline — and why your measurement instrument is probably the problem.
Originally shared on LinkedIn →