9 environmental signals this cycle
2 critical, 7 warning, 0 informational. Highest-confidence alert: temp 81.5°f — 3.5°f above flower target.
View evidence
The platform reads its own data and writes a briefing. Cohort performance, anomaly summaries, lineage integrity, market signals, research notes. Every briefing carries an explicit confidence score. When the underlying data changes, the narrative changes.
Briefings are produced by a rule-based engine that runs on every page load. The engine reads the current state of the platform (anomalies, harvests, lineage, experiments, consumer logs), ranks the most actionable signals, and writes a short narrative per briefing with an explicit confidence score.
Every page load re-reads the current platform state: anomalies, harvests, lineage graph, experiments, consumer logs.
Rule-based weighting identifies the most actionable signal across each briefing type. Confidence scored per briefing.
Each briefing writes a short narrative with the metric, the supporting data, and the suggested operational response.
Briefings appear on the Pulse landing and in this feed. When ML replaces rule-based, the output tightens; format stays.
Briefings regenerate every time the page loads. A snapshot of the daily briefing also appears at the top of the Pulse landing. Future versions will persist daily snapshots for longitudinal review.
A briefing's claim is only as good as its supporting data. Each card routes to the surface that backs the claim: anomalies to the live feed, cohort to the matrix, lineage to the graph, market to ecosystem, research to the research stream.
Briefings are derived via rule-based analysis of live platform data. Confidence reflects sample size and signal strength. Narratives regenerate on every load.
Scans every room for out-of-band sensor readings and drift, ranks by severity, surfaces top signals with suggested action.
Compares variety performance across facilities, flags leaders and laggards in yield and compound expression.
Integrity check on the DNA-barcoded library, redundant mother coverage, and passage-depth health.
Consumer report clustering, SKU velocity, regional effect preferences emerging from the commercial surface.
Auto-surfaced findings from the experiment table and observation stream, graded by confidence.
Cherry Pie averaged 117.1 g/plant across 10 harvests, outperforming Blue Dream (77.6 g/plant). THC averages 22.8% vs 22.9%.