Grade the whole cohort and the risk spreads wide — 16% top grade, 37% highest-risk, everything in between. A single vehicle never shows you that.
289 packs · live lender-facing risk grades
Location is solved. Condition — the thing that governs value, service, warranty and resale — isn't measured at all. That's the money leaking out of every EV fleet, and it's what we read.
Everyone gives you a score. We tell you what caused it — charging habits and missed maintenance together account for roughly 62% of attributed capacity loss, more than five times what calendar aging did.
Mean contributions across a cohort — they don't sum to exactly 100, and we don't force them to.
A warranty claim is an argument nobody wins. The operator points at the battery, the maker points at the route, and the relationship settles it.
The claim becomes a conversation about facts — and the share that came from operating conditions is no longer the maker's to absorb.
Settled in 90 seconds, not three weeks — with the split evidenced.
A health score called this pack mid-pack. It has nine weeks left. That gap is the whole argument.
A lender, a fleet manager, a maker and a resale desk look at one pack and need four different decisions. Same reading. Four answers. Each in your language, each with the money attached.
CAN telemetry, BMS output, telematics feeds, service records, charging logs. Whatever exists.
No minimum. Send what you have and we'll tell you what we can answer from it.
Guess. If the data can't support a finding, we say so rather than model around the gap.
API-first. We read from your systems; answers go back into them.
97M+ CAN telemetry rows processed to date.
A lender is asked to fund 200 rickshaws in a 44°C city. Everything that decides whether the loan gets repaid is a guess. Here's how we open it.
Credit, financials, a warranty sheet. The battery, untouched.
Cell divergence, measured capacity, charge pattern, 44°C thermal exposure.
31% charging, 31% maintenance, 16% usage, 11% calendar, 8% thermal.
58 km real range against 105 spec. About 9 weeks left.
Grade D on the weak packs — fund with a covenant that fires early. Deal holds.
Range, resale, reuse, recycle. Four decisions decide what a battery is worth — every one of them is a condition question.
When new batteries get expensive, the ones already on the road matter more — and all four of these are condition questions. Guess them, and you lose money at both ends.
Anyone can buy the physics. Nobody can buy 13 months of what Indian roads did to real packs. Every pack we read in a new city, duty cycle or rider profile makes the next one sharper — a better-funded rival can't shortcut this, only start counting.
Physics-driven ML on real-world data. Everyone has the physics. We have what the road did to it.
Separating cause is what makes an answer actionable.
Field data on real roads that can only be accumulated.
We see maker, operator and lender; no one else does.
Validated on real packs across 13 months and 8 cities — the actual roads your fleet runs on. Small on purpose: big enough to prove the machine works end to end, and sharper with every pack.
Talk to usAnonymised results from live platform data — one operator cohort, hundreds of packs. No names, because the findings stand on their own.
Grade the whole cohort and the risk spreads wide — 16% top grade, 37% highest-risk, everything in between. A single vehicle never shows you that.
289 packs · live lender-facing risk grades
Charging habits and missed maintenance drove roughly 62% of capacity loss in this fleet — more than five times what age and heat did.
285 packs · platform attribution engine
Real daily range ran about 21% below the spec sheet — roughly 83 km delivered against a 105 km claim, under actual duty cycles.
181 packs · measured vs manufacturer spec
Of the not-yet-critical packs flagged for cell imbalance, 98% were recommended for a balancing service, not a replacement — a recoverable problem caught early.
66 packs · platform recommendation, pre-service
packs were queued for replacement and didn't need replacing. All 14 stayed in service.
Knowing when not to act is worth more than knowing when to act.
of warning before the first missed payment — while it's still a restructuring conversation.
Condition-based early warning, not a payment prediction.
of lead time before the health floor was breached — on a pack the BMS never flagged.
Lead time is the whole product. Everything else is a report.
Anonymised from live platform data. We don't publish what we can't stand behind.
We're two founders building operational intelligence for the Indian EV battery chain — one from the commercial side, one from product and physics. We work with a small number of partners at a time. No long commitment, and a response within 48 hours.