Udyog Vihar should be prioritised for clean-air investigation
One map, six readings of the city.
A boundary, a measurement, a facility, an open investigation and a belief, read against each other on one geography.
Detail
Open →What should the room do first?
Ranked priorities from the figures on this page, each with an owning department and what would prove it wrong. Advice to an officer, not a decision.
CPCB_API_KEY to add official ground stations.Neighbourhood response pulse
2 reports received · 2 still open · 0 overdue · 0 verified RWAs
Priority investigations
Signals requiring field verification
Department action tracker
Operational accountability
Log a reading on the air view, open an investigation on the operations view, let the City Brain propose the action, then record inspection evidence.
Monitoring data.
Capture pollutant measurements with provenance and quality status.
Add a reading
Official geography.
MCG wards and colonies as published by GMDA OneMap, with the operational owner for each.
City facilities.
Places a field team may have to reach, verify or protect.
Investigate the signal.
Record anomalies, hypotheses, evidence and confidence for field verification.
Log a hotspot
Investigation register
All recorded signals · 1
Sector 37 (CAMS modelled) recorded CPCB AQI 312 (Very Poor), PM2.5 88.7 and PM10 121.5 µg/m³ at 2026-11-10 11:00. Modelled value from Open-Meteo CAMS, not a ground station. CAMS resolves to roughly a 40 km grid, so neighbouring zones can return the same figure and a local hotspot is invisible to it. Treat it as a prompt to measure, not as a measurement. Fine-fraction dominated (PM2.5 is 73% of PM10). That pattern is typical of combustion — waste or biomass burning, vehicle exhaust, DG sets. Look there first. First step: Field walk focused on burning, idling and DG sets; suggested owner MCG Sanitation / HSPCB. Confirms or rules out: A co-located ground reading plus geo-tagged field photographs. Rule out regional transport by comparing neighbouring zones over the same hours. Drafted from an Azure OpenAI briefing over this database. Verify on site before any enforcement; no source is attributed.
Confidence: Low · Status: Open · Owner: UnassignedGurugram Clean Air Brain
A living operational model that connects evidence, beliefs, decisions, interventions and measured outcomes.
Latest reasoning cycle
What the system observed and inferred
Scanned 7 zones; captured 534 evidence records; created 7 temporal zone beliefs and 0 proposals for human review.
Cognitive operating loop
From raw signals to accountable action
City world model · Current beliefs
These are provisional hypotheses, not established facts. Open the zone API for evidence lineage.
Vikas Sadan should be prioritised for clean-air investigation
Golf Course Extension should be prioritised for clean-air investigation
Manesar should be prioritised for clean-air investigation
NH-48 should be prioritised for clean-air investigation
Sector 37 should be prioritised for clean-air investigation
TERI Gram should be prioritised for clean-air investigation
Decision queue
Brain-generated proposals. Nothing becomes a work order without a named human decision.
Inspect DG set operating and emission controls
Most frequent citizen-observed category in the last 30 days: DG set emissions / noise (1 reports). Validate on site before enforcement.
Decide
Conduct a geo-tagged burn-site inspection and repeat check
Most frequent citizen-observed category in the last 30 days: Open garbage burning (1 reports). Validate on site before enforcement.
Decide
Run a coordinated hotspot source-verification visit
this signal prioritises investigation but does not identify a source.
Latest recorded AQI is 207
Decide
Latest recorded AQI is 207
Decide
Latest recorded AQI is 207
Decide
Latest recorded AQI is 207
Decide
Latest recorded AQI is 212
Decide
Latest recorded AQI is 207
Decide
Latest recorded AQI is 207
Decide
Evidence ledger
Source, time, quality and provenance retained for every signal.
A mapped asset is a place to verify, not a finding and not an emission source. Facility presence is never evidence of pollution. Nothing here has been field-verified — every asset was bulk-imported and the app has no inspection step that could mark one verified, so the right-hand column shows where the record came from instead, which is what should change how far you trust it.
Operational memory
Reviewed lessons from measured interventions.
A lesson is only recorded after an approved action has been carried out and measured. The system does not learn silently from unreviewed outcomes.
Outcome capture · Close the learning loop
Record a measured outcome against an approved proposal, then promote a reviewed lesson.
An outcome is recorded against an approved proposal once its work order has been carried out and measured. Approve something in the queue above first.