Your technician photographs an extinguisher or a hose reel. The engine returns a graded finding: critical, major or minor, each one cited to IS 2190, IS 15683 or NBC 2016 Part 4, with the Maharashtra Rules 2009 overlay. Thirteen checks, and the severity of each one is fixed in code rather than decided by the model.
Most vendors describe their checks in adjectives. Here is the actual list, with the severity we assign and the clause each one answers to. The severity is authored in code, not left to the model, which is what makes two audits of the same unit comparable, and what lets you argue with a grading decision instead of shrugging at it.
| Check | Severity | What it catches | Reference |
|---|---|---|---|
| UNIT_MISSING_OR_DISCHARGED | CRITICAL | Absent, discharged, or gauge sitting in the recharge zone | IS 2190; NBC 2016 Part 4 |
| ACCESS_BLOCKED | CRITICAL | Access fully blocked, or the unit obscured | IS 2190 cl. 4; NBC 2016 Part 4 |
| HOSE_REEL_UNUSABLE | CRITICAL | Hose missing or severed, cabinet will not open | NBC 2016 Part 4 |
| GAUGE_OUT_OF_RANGE | MAJOR | Needle outside the green band, or an unreadable gauge face | IS 2190 |
| SEAL_OR_PIN_COMPROMISED | MAJOR | Pin missing or unseated, tamper seal broken | IS 2190 |
| REFILL_OR_EXPIRY_OVERDUE | MAJOR | Refill or expiry date passed | IS 2190 cl. 7 |
| INSPECTION_TAG_MISSING | MAJOR | Tag absent, illegible, unsigned or stale | IS 2190; MFPLSM Rules 2009 |
| PHYSICAL_DAMAGE_OR_CORROSION | MAJOR | Corrosion, dents, damaged hose, horn or nozzle | IS 2190 |
| MOUNTING_HEIGHT_WRONG | MAJOR | Insecure, or top above ~1.5 m for a hand-portable unit | IS 2190 cl. 4 |
| WRONG_CLASS_FOR_HAZARD | MAJOR | e.g. a water-based unit at an energised electrical hazard | IS 2190 cl. 3; NBC 2016 Part 4 |
| ISI_MARK_MISSING | MINOR | ISI / BIS mark absent, painted over or illegible | IS 15683; BIS certification |
| CABINET_DEFECT | MINOR | Glass, latch or break-glass defects; equipment still usable | NBC 2016 Part 4 |
| SIGNAGE_MISSING | MINOR | No location marking where the unit is not plainly visible | NBC 2016 Part 4 |
The severity boundary is a judgement, and we state it so it can be argued with. Critical means the equipment cannot be relied on to work, or cannot be reached. Major means a defect that defeats certification or maintenance requirements. Minor means marking, documentation or housekeeping.
The verdict itself: pass, fail or re-inspect, is derived in code from those severities and stored alongside them. So the stored result is reproducible from the stored evidence: you can audit the audit. And it looks for the ISI mark, not UL or FM. An engine tuned for American equipment will cheerfully report that your extinguisher has no UL listing, which is technically true and completely useless to you.
Stated plainly, because the fastest way to lose your trust is to let you discover a limit after you have paid for it.
The technical basis is national: IS 2190, IS 15683 and NBC 2016 Part 4 apply across India. What is specific to Maharashtra is the statutory overlay: the Maharashtra Fire Prevention and Life Safety Measures Act 2006 and Rules 2009, and the authority context of the Chief Fire Officer of the municipal corporation. MCGM for Brihanmumbai. We have built and tested it against that context first. Taking it to another state is a change to the statutory overlay, not a rebuild of the engine.
Let us be precise about what this does and does not replace. It does not remove your technician, and it does not remove the qualified person who signs off. What it removes is the report production between them: transcribing findings, finding the right clause, grading each one, and formatting it for review.
The premises takes a morning. The file takes longer, transcribing, hunting the clause, grading each finding, typing it up for signature.
A discharged unit is seen on Monday and written up on Thursday. The defect sits in a notebook while the paperwork catches up.
Grading by hand drifts between people and between visits, and inconsistency is exactly what an inspection challenge attacks.
Photograph to graded verdict in seconds, filed with its evidence. A critical finding alerts an operator immediately.
The same qualified person still reads every finding and still signs, but reviewing a drafted, clause-cited verdict takes minutes where writing one took hours.
Severity is fixed in code and the verdict derived from it, so grading does not drift between technicians, premises or half-years.
The only figure we assume below is the ~15 minutes it takes to read and approve a drafted, clause-cited verdict. The rest is yours: find your own write-up time on the left, and swap in your own monthly volume.
| If your write-up takes today… | You get back, per report | Per 40 reports filed |
|---|---|---|
| 1 hour | ~45 minutes | ~30 hours |
| 2 hours | ~1¾ hours | ~70 hours |
| 3 hours | ~2¾ hours | ~110 hours |
We stop at hours on purpose. What an hour of your capacity is worth, absorbed internally or rebilled, is your figure, not ours. Bring it to the call with your real write-up time and volume, and we will run the payback on your actual numbers instead of our assumptions.
The hour count is the easy part to model. The reason a firm keeps it is usually the other half: every finding cites the clause it fails and its severity is fixed in code, so the same photograph produces the same grade in July as it did in January: across technicians, premises and half-years. Hand grading drifts, and drift is what a challenge from the fire authority, an insurer query or a disputed finding attacks first.
When that challenge comes, the answer is a record rather than a recollection: an append-only log, the evidence photograph retained against the finding, the clause quoted, the timestamp, and the name of the qualified person who approved it. The stored verdict is reproducible from the stored evidence, you can audit the audit.
Pricing below is specific to the AI Field Auditor only, other systems are scoped separately.
Start on the monthly. It is the smallest commitment that still puts the engine on your own premises and your own equipment, it is an operating expense rather than a capital sanction, and there is no infrastructure of yours to maintain. Owning it outright is what you graduate into once the throughput is proven on your own work.
| Engagement | Best for | Investment |
|---|---|---|
| Launch Pilot | Prove it on your own premises and your own jurisdiction before committing to anything monthly. One-time offer per business. | from ₹1,50,000 one-time |
| Managed Subscription most firms start here | We host, run and keep improving it, an operating expense instead of a capital sanction, and no server of your own to run. | Starter ₹34,999 / month 500 audits / month Pro ₹69,999 / month 1,000 audits / month |
| Build & Own It | Own it outright on your own infrastructure, far cheaper over its life, the records never leave your estate, and payable across build milestones rather than as a single cheque. | ₹15,00,000 one-time special introductory offer, ≈ 21 months of the Pro monthly, then yours |
Put the outright figure in proportion before it frightens anybody. At ₹69,999 a month, a managed client's cumulative spend passes ₹15,00,000 at about 21 months. So owning it is roughly under two years of what you would be paying anyway, except that at the end of it the engine is yours, on your own infrastructure, with nothing further to pay. A firm that inspects for many years never stops paying on the monthly; that, and not the sticker, is the argument for owning it.
And it is not a single cheque. Build & Own It is payable across build milestones: scoping, configuration against your equipment classes and premises naming, and handover, so the spend tracks delivered work rather than landing up front. One audit is one photograph judged; a submission rejected before it is judged costs you nothing and does not count.
In Maharashtra, certification is the business of a Licensed Agency, licensed by the Director of Maharashtra Fire Services, and the signing engineer's standing comes from qualification. That is not an obstacle to this engine; it is the shape of it. The engine is advisory by architecture and always requires a qualified sign-off, which means it is built to sit under your licence, not against it: white-labelled output, your letterhead, your premises naming, your engineer's approval on the record, and on an owned deployment the evidence never leaves your estate.
If that is your position, say so on the call, a per-report wholesale arrangement is usually a better fit than either subscription tier, and we scope it against your volume and premises mix.
To be unambiguous: this is roadmap, not a feature you would be buying today. The engine currently produces advisory screening, and every response says so.
But it is worth telling you where the design is headed, because it is the reason we built the Indian engine to derive its verdict in code rather than accept whatever a model returned. Under the Maharashtra Act, fire safety compliance is evidenced through Form B, prepared half-yearly. January and July. That work is largely done out of files and spreadsheets.
A system that accumulates a premises' audits across the half-year and assembles the supporting evidence pack for that period is solving a real, dated, recurring obligation rather than offering a nicer inspection form. That is the next phase. It needs a signature to mean something first, which is why the verdict is now reproducible from the stored evidence, and why sign-off was deliberately not built until that was true.
If the Form B cycle is the problem you actually want solved, tell us on the call. It moves up the queue when a real premises is waiting for it, and we would rather build it against your half-year than our assumptions.
The engine produces a documented, clause-cited judgement and stops. A qualified person reviews and signs off. It is not a Form B certificate and not a substitute for a Licensed Agency, that is the design, not a limitation of a cheaper tier.
It does not classify equipment, it assumes what it is shown is an extinguisher or a hose reel. Sprinklers, detectors, panels, pumps and doors are not covered by this build. We would rather tell you that here than have you find it in week three.
A clear, well-framed photograph yields the best result. When the evidence will not support a judgement the engine returns re-inspect rather than guessing. You will see that verdict, and you should want to.
It looks for the ISI / BIS mark and cites IS 2190, IS 15683 and NBC 2016 Part 4. It is explicitly told not to reason from NFPA, IFC or UL/FM. Our US build is a separate system for a separate code basis.
The engine and the dashboard are live. Bring a photograph of one of your own extinguishers to the call and watch it become a graded, clause-cited finding in front of you, or send us ten from a recent round and we will send the report back within 24 hours.