02/LabExperiments · prototypes · notes
I'm always building something.
A workbench for ideas that are not products yet: agents, models, game prototypes, tools and architecture tests. Some become systems. Most become lessons.
- Experiments
- 006
- Active
- 006
> next experiment loading
- EXP.001
Multi-agent task routing
Agents negotiating who owns a task before any code is written.
RunningAI Agents - EXP.002
Local LLM evaluation harness
Repeatable evals for prompts and tools, run on every change.
RunningLLMs - EXP.003
Procedural terrain sketches
Noise fields and contour lines — the terrain on this page is one.
RunningThree.js / Canvas - EXP.004
Crafting graph prototype
Recipes as a dependency graph instead of a lookup table.
RunningUnreal Engine - EXP.005
Dev environment snapshots
Disposable, reproducible workspaces for humans and agents.
RunningDeveloper Tools - EXP.006
Event-driven service mesh
RabbitMQ, Redis streams and the cost of eventual consistency.
RunningArchitecture