KshatiPro — damage-estimate writer for independent Indian motor surveyors

1. One-liner

Turns a surveyor’s crash-site photos into a line-item repair estimate and a formatted IRDAI motor survey report.

2. Trend signal — why now?

Three things converged in the last 12 months, and I can point at all three.

The work is still done by hand. On the licensed-surveyor forums, the consensus is blunt: “Mostly surveyor brothers work on their own format either in word or excel. Its simple and costless” (Naresh Kukkar, insurance_surveyors Google Group). A licensed IRDAI motor surveyor inspects a damaged car, photographs every panel, then hand-types each part, labour line, depreciation, and salvage figure into a Word or Excel template. The #1 complaint isn’t the assessment — it’s the drudgery: clumsy tables, borders, headers, and computers “flooding with unmanageable Album, Photos, Word, Excel and PDF files.”

The legacy tools are formatting tools, not brains. Code-X “Survey Solution” (deployed in 21–23 states) and SurveyorLite are real, paid, and adopted — but they’re album-builders, depreciation calculators, and report formatters. Confirmed directly: SurveyorLite is “a workflow and valuation management platform rather than an AI-powered damage estimator.” Nobody is reading the photo and drafting the estimate. That’s the entire manual core still left on the table.

The multimodal capability that closes that gap is now cheap. 2025–26 vision LLMs can look at a dented quarter-panel, identify the part, infer repair-vs-replace, and draft a line-item estimate against a parts/labour reference — a task that needed a human eyeball 18 months ago. One incumbent already bolts “AI import from PDFs/images” onto its formatter, proving the plumbing works; none has made it the product.

And the regulator is pushing turnaround. IRDAI’s TAT regime penalises insurers (penal interest + daily fines) for slow settlements; the surveyor report is the bottleneck (inspection within 48–72 hrs, report right after). Insurers push that pressure down onto the independent surveyors who do the volume.

Provenance:

• Signal 1 (Demand): Surveyors work in hand-built Word/Excel; top complaint is formatting drudgery and file chaos; verbatim forum quotes — https://groups.google.com/g/insurance_surveyors/c/kfKeUgPJbbU and http://code-x.in/Survey_Solution.html — 2026-07-26
• Signal 2 (Feasibility): Existing paid tools (Code-X Survey Solution, SurveyorLite ₹20/survey) are formatters/calculators with NO AI damage estimation; multimodal LLMs now draft line-item estimates from photos — https://www.surveyorlite.com/ — 2026-07-26
• Signal 3 (Economic): IRDAI TAT penalties on insurers push report turnaround; India insurance sector heading to ~USD 222B by 2026; IRDAI moving surveyors to perpetual registration — https://taxguru.in/corporate-law/irdai-proposes-perpetual-registration-insurance-surveyors-loss-assessors.html — 2026-07-26 Category: Geographic arbitrage

3. The opportunity

The manual heart of a motor survey — look at damage, decide repair vs replace, write the line-item estimate — is the one thing every Indian surveyor still does by hand, and the one thing no existing tool does for them. The incumbents (Code-X, SurveyorLite, MSMS) fight over formatting, albums, and depreciation math. That’s the packaging of the report. The content is wide open.

KshatiPro is an AI-first take: the surveyor shoots the panels on their phone, the model returns a draft line-item estimate (part, repair/replace, labour, paint), and the surveyor edits and signs off. The report and photo album format themselves as a byproduct. The disruption isn’t “prettier reports” — it’s collapsing the 45–90 minutes of manual write-up per claim into a 5-minute review.

This is a geographic-arbitrage play: the US has Xactimate/CCC as a $1B+ estimating standard; India has surveyors typing into Excel. The wedge is bringing the estimate-generation layer — not the enterprise claims platform — to the independent professional.

4. Target market

• Primary customer: Independent IRDAI-licensed motor insurance surveyors & loss assessors — solo practitioners and 2–5 person surveyor firms doing motor own-damage (OD) claims. Not the top-6-insurer enterprise buyers (that’s Roadzens/xClaim’s game).
• Why they buy: In their words — “editing is not possible… net is compulsory to run the software” and “much-much higher side” on price; they’ll pay only if it saves real time, not for another formatter. The pull is: stop hand-typing 40 part lines per claim during peak season.
• Rough TAM reasoning: Tens of thousands of licensed surveyors nationwide (IRDAI register), the majority handling motor/minor-property volume. Even a few thousand active independents doing OD claims is a serviceable base. Realistic obtainable niche: 1,000–3,000 paying surveyors.
• Why now for them: Motor claim volumes rising; IRDAI TAT penalties flow downstream as turnaround pressure; the ₹75,000 motor threshold means high claim frequency needs a surveyor. Faster write-up = more claims/month = more fee income (fees are 0.5–2% of assessed loss).

5. Product sketch (MVP)

• Phone capture: shoot damaged panels; app tags each photo to a panel/part.
• AI draft estimate: model returns line-item repair/replace list with labour and paint, against an editable Indian parts/labour reference.
• One-tap depreciation, salvage, policy-excess, and total-loss flags (table-stakes to match SurveyorLite).
• Auto-built photo album/sheet + IRDAI-style spot/interim/final report in Word/PDF, using the surveyor’s own template.
• Full inline editing (directly answering the #1 forum complaint that incumbents can’t be edited).
• Works with patchy connectivity — capture offline, sync later (kills the “net is compulsory” objection).
• Hindi + regional-language UI and remarks library.
• Per-claim history that’s actually searchable (another named gap: “you can not search on the report when required”).

6. AI angle — what’s load-bearing

The AI is the product. Remove it and you’re left with SurveyorLite — a calculator and formatter that already exists and that surveyors grudgingly pay ₹20/survey for. The load-bearing work is the vision model reading damage photos and drafting the line-item estimate: the single manual task no incumbent automates. Everything else (depreciation math, album layout) is commodity glue around that core.

7. Localization angle

This is India-native by construction, not a translated global tool:

• Report format: IRDAI spot/interim/final report structure, Form 12, GST fee bills — not Xactimate’s US format.
• Parts/labour reference: Indian OEM part nomenclature and local labour rates, not US price lists.
• Language: Hindi + regional UI and a vernacular remarks library; surveyors work small-town claims.
• Pricing rail: UPI, per-survey micro-pricing (₹/survey) matched to how they already buy (SurveyorLite bills ₹20/survey).
• Connectivity: offline-first capture for weak rural data.

A generic global estimator cannot serve this buyer. That’s the moat’s foundation and the reason a US player won’t bother.

8. Business model — path to $1M–$5M ARR

• Pricing: Freemium-to-paid, ₹/survey to match buying behaviour. ~₹40–60 per completed AI estimate, or ₹800–1,200/mo unlimited for high-volume surveyors. (Benchmark: SurveyorLite ₹20/survey, ₹10,000/yr premium; legacy MSMS ₹250/mo.)
• ACV: ₹8,000–12,000/year per active surveyor ($95–145).
• Rough math to $1M ARR (~₹8.3 Cr): ~7,000 surveyors × ₹12,000/yr. That’s a large slice of the active-independent base — aggressive but not fantasy given tens of thousands licensed.
• Rough math to $5M ARR: requires expanding beyond motor into property/marine survey estimates, plus a possible insurer-side channel (insurers paying to receive faster, structured reports) — a different, harder motion.
• Expansion path: motor → property/fire → an insurer-facing “structured report intake” upsell where the carrier pays for standardized, faster submissions.

9. Go-to-market wedge — first 100 customers

• The surveyor associations and groups already exist and are reachable. IIISLA chapters, the insurance_surveyors Google Group, and state surveyor WhatsApp/Telegram groups are where these exact people argue about software today. Post a 90-second before/after video (Excel typing vs 5-min AI draft) into the groups where the ₹250/mo complaints happened.
• Free-tier land grab: first 20 estimates free (mirrors SurveyorLite’s 20-free hook), no card. Cheap buyers try free tools.
• City-by-city ambassador: recruit 1 respected senior surveyor per metro (Mumbai, Delhi, Pune, Hyderabad, Ahmedabad) to demo in their IIISLA chapter for a referral cut.
• Peak-season timing: launch outreach into monsoon/festival accident-spike months when write-up backlog is worst and time pain is highest.
• Direct cold outreach: IRDAI’s surveyor register is public; contact independents doing motor lines directly with a personalized sample estimate generated from a demo photo.

10. Build complexity — justification

Medium. The vision-to-estimate model on off-the-shelf multimodal APIs plus a report/album generator is standard AI-app work. The genuinely hard parts are (a) building a credible Indian parts/labour reference so the draft is trustworthy, and (b) getting damage-assessment accuracy high enough that editing beats typing from scratch — this needs a domain-expert surveyor in the loop from day one. Small team, ~3–4 months to a usable v1.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Selling a drafting tool to already-licensed surveyors; surveyor signs & owns the report. No IRDAI product approval needed.
Ethical — no harm / dark patterns	✅	Human-in-the-loop; surveyor reviews and signs. Must guard against over-trusting AND under-scoping estimates.
Market exists (evidence above)	✅	Paid incumbents, forum demand, verbatim complaints.
1–5 person team can build this	✅	Off-the-shelf multimodal + report gen + reference data.
Launchable with <$50K / ₹40L	✅	Solo/small team, API-based, no hardware.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	14/20	Real, felt regularly — but they’ve built cheap workarounds (Word/Excel, Dropbox) and call even ₹250/mo “high side.” Painful, not hair-on-fire.
Demand evidence	15	10/15	Multiple paid incumbents + verbatim forum complaints = solid signal, but the AI-estimate specific demand is inferred, and some challengers (SurveyorLite, FieldNotes voice-to-report) are already circling.
Build feasibility	15	10/15	Standard AI plumbing, but accuracy bar for “edit beats retype” plus building a trustworthy Indian parts/labour reference is real work.
Distribution clarity	15	10/15	Named channels (IIISLA, surveyor Google/WhatsApp groups, public register) — but the audience is small, skeptical, and cheap.
Revenue mechanics	15	8/15	Pricing benchmarked and clear, but WTP ceiling is low (₹20/survey resistance) and ACV is thin; $1M ARR needs a big slice of a small base.
Time to first revenue	10	7/10	Free-to-paid funnel into existing groups; realistic first paid users in 6–8 weeks post-MVP.
Defensibility	10	5/10	Parts/labour reference + workflow lock-in accrue over time, but core is copyable and AI-native challengers are appearing. Execution moat, not structural.
Total	100	64/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

Key assumptions to validate (3–5)

1. Assumption: AI draft estimates are accurate enough that editing is faster than typing from scratch. How to test: run 50 real damaged-car photo sets past 5 working surveyors; measure edit-time vs their manual write-up time. Need ≥50% time saving to matter.
2. Assumption: Surveyors will pay ₹800–1,200/mo (4–5× the ₹250 legacy tools) because the AI saves real time, not just formatting. How to test: pre-sell to 30 surveyors at target price after they try it on 5 live claims.
3. Assumption: The Indian parts/labour reference can be assembled cheaply enough to make drafts trustworthy. How to test: build the reference for 3 common models (Swift, WagonR, i20) and measure draft accuracy before scaling.
4. Assumption: Association/group channels convert. How to test: post the before/after demo in 3 surveyor groups; measure trial signups per 100 members.

Risk flags

1. Cheap, skeptical buyers: the loudest recorded opinions are “much-much higher side” and “wastage of precious time.” This audience resists paying for software. Biggest threat to revenue mechanics.
2. Incumbent + AI-challenger squeeze: legacy formatters (Code-X, MSMS) own habit; new AI-flavored entrants (SurveyorLite adding AI import, FieldNotes voice-to-report) are moving into the same lane. Window is narrowing.
3. Accuracy/liability: a bad AI estimate that a surveyor signs could mis-settle a claim. Must stay firmly human-in-the-loop and never auto-submit.
4. Small TAM at low ACV: the math to $1M ARR requires converting a meaningful fraction of a modest base — attractive as a lifestyle/bootstrapped business, tight as a scale play.

14. Structured verdict

Score:                  64/100
Verdict:                VALIDATE
Confidence:             Medium
Best-fit builder:       Technical founder + a working IRDAI motor surveyor as co-founder/advisor
Time to revenue:        6–10 weeks after a 3–4 month MVP
Capital to launch:      ₹8–15 lakh ($10–18K)
Top 3 assumptions to validate first:
  1. AI draft accuracy — 50 real photo sets, edit-time vs manual write-up, need ≥50% time saving
  2. Willingness to pay 4–5× legacy price — pre-sell 30 surveyors at ₹800–1,200/mo after live trial
  3. Parts/labour reference is buildable cheaply — 3 common models, measure draft accuracy
Kill criteria:
  - Abandon if <50% of trial surveyors report time saved vs their manual Excel process
  - Abandon if <10% of free-tier trials convert to paid at target price after 2 months
  - Abandon if a funded AI-native entrant locks the surveyor associations before MVP ships

15. Next step — 1-week validation sprint

• Day 1–2: Collect 30–50 real damaged-vehicle photo sets (from a friendly surveyor’s past claims). Run them through an off-the-shelf multimodal model with a hand-built parts/labour prompt. Grade the draft estimates against the surveyor’s actual reports.
• Day 3–4: Sit with 5 working surveyors. Have them edit the AI drafts into final reports. Stopwatch it against their normal Excel write-up time. Ask the money question: “would you pay ₹1,000/mo for this?”
• Day 5: Go/no-go. Go only if ≥50% median time saving AND ≥15 of the surveyors say yes to ₹1,000/mo. Otherwise the price ceiling kills it — park it as a VALIDATE and revisit when model accuracy or the surveyor base shifts.

Falsifiable outcome: measured edit-time saving and a hard count of pre-commit “yes at ₹1,000/mo.” Not vibes.