Policy tailwind — RDSS
The Revamped Distribution Sector Scheme is the Government of India's ₹3.03 lakh crore distribution programme, sanctioned 2021, running to FY 2025–26. Its smart-metering component targets ~250 million prepaid smart meters plus feeder and distribution-transformer metering — the exact measurement layer this system consumes.
2021Scheme sanctioned
2023Meter tenders at scale
2025–26Scheme end-date
Post-26DT-level optimisation
Public, checkable milestones — stated so the claim can be verified rather than taken on trust. We are not affiliated with the scheme; it is cited as market context for why per-flat and per-transformer data is becoming available.
Operations view · mockup
Sub-meters reporting
--
across the district
Coordinators active
--
one per society
These two counters are derived from the loaded district datasets, but the framing — a production monitoring console — is a mockup of what operations would look like at scale. No such console has been built.
This colony
Three synthetic societies on a shared feeder — Society A (60 households, the detailed one), Society B (48, an older/smaller transformer, runs hot), Society C (48, a newer better-rated transformer, has headroom). Fully synthetic: archetype behaviour and appliance ratings are hand-authored, not fitted to a named dataset. Nothing here is claimed as real-time.
Each house's sub-meter
Every house's Load now is its own metered reading — the same granularity a real per-flat sub-meter reports, and the same data path GridBrain's decision ledger reads from directly. No discom API involved; this is a bulk HT society's own billing sub-meters.
What GridBrain actually does here
The curtailment lines in the ledger are the real, already-simulated engine decisions (the same transformer-protection logic that runs in the physics), surfaced in plain English as they happen — not a separate fabricated demo layer. The cross-society "headroom" lines are a live computed comparison across the three transformers, framed as a recommendation: this colony does not yet simulate an automated cross-feeder transfer mechanism, and that's stated here rather than implied.
The sibling single-building benchmark
Aethergrid's forecasting method (95th-percentile chance-constrained dispatch) is validated separately on real meter data: Building Data Genome Project 2 — four real buildings, two years, held-out test months, American sites (Tempe, Arizona), openly disclosed. Only the method transfers, not the load shapes. Tariff, bill engine and calibration are checked against a real published Tamil Nadu (TNERC) order, at aethergrid-vert.vercel.app/method — a different, complementary piece of evidence, not this same system.