An evidence engine for agent work.

BrainSNN brings content analysis, reviewed draft improvements, browser-local scan history and proof missions into one product. Start with a passage, page or screen recording, inspect the evidence, and make a change you can test.

The public detector evaluation reports results on unseen passages, including missed cues and false alarms. Model scores are review signals, not measurements of a reader’s brain or predictions of market outcomes.

The engine roadmap connects versioned, reusable context with independent tests and evidence-based resource allocation. This is planned integration, not a claim of autonomous learning already running.

The office preserves a recorded window into XIO’s separate operational experiment. Its approved counts and events are not BrainSNN-wide engine metrics. Unknown results remain unavailable. Reconstruct, the Arcade and neural circuit simulations remain available as tools and research.