SB StartupBasket
All ideas
75 /100 GO Medium complexity

OpnameLock — evidence ledger for Indonesian subcontractors

Turns a subcontractor's daily site photos and voice notes into an owner-acknowledged, tamper-evident progress record that gets the termin paid.

TradeTech SaaS SEA WhatsApp-firstVoice-firstMultilingualAI-agentSMBMobile-first
— views
Evaluation Scores
75/100

GO

Overall Score

17
Problem
12
Demand
11
Build
11
Distrib.
11
Revenue
7
Time
6
Defense

OpnameLock

1. One-liner

Turns a subcontractor’s daily site photos and voice notes into an owner-acknowledged, tamper-evident progress record that gets the termin paid.

2. Trend signal — why now?

Indonesia’s construction-subcontractor payment crisis is not a hunch — it’s documented across legal, academic, and news sources, and a fresh AI capability finally makes the defensive tool buildable on a ₹-cheap phone.

  • Demand (severe). Gapensi (the national construction-executors association) recorded 37,000 private contractor bankruptcies in three years, the cited reason being lack of work or delayed payment after work was completed (Kaskus/Gapensi thread). Named 2026 cases: subcontractors unpaid on IKN (new capital) projects and by Adhi Karya subsidiaries; an April 2026 vendor protest over a Rp3.9B Ministry contract (Poskota TV, 16 Apr 2026).
  • Root cause is documentation, not just cash. Indonesian law firms list the exact evidence an unpaid subcon needs — subcontract copy, Berita Acara Serah Terima Pekerjaan, official invoice, payment-request correspondence, somasi proof — and warn that delay risks “potensi kehilangan bukti administratif” (loss of administrative evidence) (ILS Law Firm). The ASCE Journal of Legal Affairs paper on conditional payments confirms “administrative disputes regarding work documents” as a recurring failure mode (ASCE).
  • Feasibility unlock (last 12 months). Cheap multimodal vision-language models now read a phone photo plus a Bahasa voice note and emit structured progress data. The enterprise vision incumbents (DroneDeploy Progress AI launched Jul 2025, OpenSpace, Buildots, Doxel) prove the technique works — but they require drones, 360 rigs, and 4D BIM, and price for US/EU GCs (DroneDeploy). The unlock here is that none of that hardware is needed for a defensive evidence product.
  • Economic. Indonesia construction-tech funding roughly doubled YoY into 2026 (Tracxn lists 44 constructiontech startups, 12 funded) — but every funded player (Manpro, SipilPro, PROGRESI, HashMicro) builds main-contractor / owner project-management suites. The subcontractor’s defensive side is unserved (Tracxn).

Provenance:

3. The opportunity

Indonesian construction runs on pay-when-paid / pay-when-certified. The small subcontractor does the work, then waits on a berita acara opname (joint site-inspection report) and a Berita Acara Serah Terima before any termin is released. When the main contractor’s own cash dries up — or wants to squeeze — the cheapest move is to dispute the volume, withhold the signature, or simply stall. The subcontractor, who avoided pushing for paperwork to “keep the relationship” (the ASCE paper’s #1 root cause), now has no independent, dated, acknowledged record of what was actually built. Five years later the claim is statute-barred and the company is one of the 37,000.

Every incumbent app digitizes the main contractor’s daily report — i.e., the record controlled by the party with the leverage. Nobody builds the subcontractor a record that is independent of the main contractor and still credible to the owner and a court. That asymmetry is the gap. AI is load-bearing because the subcontractor’s foreman won’t fill forms — he’ll snap photos and talk. Turning that mess into a structured, quantified, owner-pingable evidence trail is the entire product.

4. Target market

  • Primary customer: Owner-operators of subkontraktor spesialis SMEs in Indonesia — formwork, rebar, MEP, finishing, earthworks crews — typically 10–80 workers, 2–8 concurrent projects, Rp2–30B annual turnover, run off WhatsApp and a site foreman (mandor). Tier-1 metros first (Jakarta, Surabaya, Medan, Balikpapan/IKN corridor).
  • Why they buy (their words): “Kami sudah kerja, tapi berita acara belum ditandatangani, terminnya nggak cair.” They have lived the unpaid-after-completion story or watched a peer go bankrupt from it. They don’t want project management — they want to get paid and to have proof if it goes bad.
  • Rough TAM reasoning: Indonesia has tens of thousands of registered specialist subcontractors plus a long informal tail; Gapensi’s 37k-bankruptcy figure implies a six-figure population of exposed firms. Capturing even 3,000 paying SMEs at Rp1.5M/mo is Rp54B ($3.3M) ARR — comfortably inside the sub-$5M target.
  • Why now for them: IKN and post-2024 infrastructure spend pulled thousands of small subs into projects with long, contested payment chains; the bankruptcy headlines have made “get it in writing” a felt priority, not an abstraction.

5. Product sketch (MVP)

  • WhatsApp/Android capture: foreman sends site photos + a Bahasa voice note (“hari ini cor kolom lantai 3 zona B selesai”) at end of day.
  • AI converts each entry into a structured daily progress record: work item, location/zone, estimated quantity/% complete, GPS, timestamp, weather, photo set — written in the contract’s bill-of-quantities language.
  • Tamper-evident ledger: each entry is hash-chained and time-anchored at capture; later edits are visible. This is what makes it credible vs. a folder of phone photos.
  • Owner/MK acknowledgement loop: a one-tap WhatsApp link to the project owner’s rep or manajemen konstruksi to “acknowledge received” (not approve) — converting a unilateral log into an acknowledged record without needing the main contractor’s cooperation.
  • Auto-assembled opname / termin claim packet: cumulative progress vs. BoQ, photo evidence index, and a draft Berita Acara Opname / payment-request letter in standard Indonesian format.
  • One-click dispute dossier: when a termin is withheld, export a chronological, hash-verified evidence bundle (+ draft somasi) ready for a lawyer or mediation.
  • Multi-project dashboard for the subcon owner: which termins are due, acknowledged, overdue, at risk.

6. AI angle — what’s load-bearing

Remove the AI and you have a photo folder — exactly the useless artifact subcontractors already have. The product exists only because a multimodal model can (a) turn an unstructured Bahasa voice note + grease-and-mud site photos into a structured, BoQ-aligned quantity/percentage claim a quantity surveyor would recognize, and (b) do it cheaply enough to run every day on every crew on a low-end phone. Vision-language progress estimation from a single handheld photo (no drone, no BIM) is the capability that did not exist affordably 12–18 months ago. The defensibility-grade part — hash-chaining + owner acknowledgement — is plumbing, but it’s worthless without the AI doing the structuring that makes the record mean something.

7. Localization angle

This is an Indonesia-first play and localization is the wedge, not a coat of paint:

  • Language: Bahasa Indonesia voice + the regional/site argot mandors actually use; output in the berita acara / opname / termin document conventions Indonesian owners and courts expect.
  • Channel: WhatsApp-first capture — the only interface a site mandor will actually use.
  • Legal format: Output maps to Berita Acara Serah Terima Pekerjaan, somasi, and the evidence list Indonesian law firms publish for wanprestasi claims. A generic global “construction progress” app produces nothing a Pengadilan Negeri or a Jakarta mediator recognizes.
  • Pricing: Rp1.5M/mo (~$90) works where a $300/mo US tool is a non-starter for a Rp2–30B-turnover sub.

A generic global product cannot win here; the entire value is in being Indonesian-document-native.

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

  • Pricing: Rp1,500,000 / month per company (≈ $90) for up to 5 active projects; Rp250,000 / extra project. Annual prepay discount to lock cash and retention.
  • ACV: Rp18M ($1,100) base; multi-project subs land at Rp25–35M.
  • Rough math to $1M ARR: ~1,000 paying subcontractors × ~$90/mo × 12 ≈ $1.08M.
  • Rough math to $5M ARR: ~3,500 subs at blended ~$120/mo (project expansion + dispute-dossier upsell). Requires winning 2–3 metros and a referral loop, not a category miracle.
  • Expansion path: per-project expansion as the sub grows; paid “dispute dossier + somasi drafting” event fee (Rp2–5M per incident — high willingness-to-pay precisely when a termin is stuck); later, a lender/insurer channel that prices receivables using the verified ledger.

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

  • Gapensi / Aspeknas chapter lists. Provincial construction-association chapters publish member directories. Scrape Jakarta + Surabaya + Balikpapan specialist-subcon members (~2,000 firms); send a Bahasa WhatsApp + 90-second screen-capture demo built around their nightmare (“termin nggak cair?”). Expect 3–5% to a paid pilot.
  • Ride the unpaid-subcon news cycle. Comment-thread and DM the named, publicly aggrieved subs in IKN / Adhi-Karya / Ministry-vendor stories and the Kaskus/Gapensi bankruptcy threads — these are people actively bleeding from this exact problem this month.
  • Construction-law firms as referrers. Firms like ILS publish “subkontraktor tidak dibayar” guides as lead-gen. Offer them a co-branded dispute-dossier export; they refer subs upstream (prevention) and we feed them clean evidence (their billable work).
  • Material-supplier & mandor WhatsApp groups. Rebar/formwork supplier reps and mandor coordinators sit in dozens of regional WhatsApp groups; a referral fee turns them into a distribution layer into exactly the target firms.

10. Build complexity — justification

Medium. Off-the-shelf: multimodal LLM/vision APIs, WhatsApp Business API, standard mobile + web stack, managed hash-chain/time-anchor. Custom work is the hard part but bounded: a reliable BoQ-aligned progress-extraction pipeline (prompt + eval harness + correction UI so a sub can fix a misread quantity), the acknowledgement-loop UX, and Indonesian document templating with a construction-law advisor. A focused pair ships a credible v1 in ~4–5 months; the eval/accuracy loop is the schedule risk, not the plumbing.

11. Gating checklist

GatePass?Note
Legal in target market✅Record-keeping/evidence tool; we draft templates, not practice law (route disputes to partner firms).
Ethical — no harm / dark patterns✅Tilts power toward the under-leveraged party; acknowledgement loop is opt-in, not coercive.
Market exists (evidence above)✅37k bankruptcies, named 2026 cases, legal+academic corroboration.
1–5 person team can build this✅Pair + part-time construction-law advisor.
Launchable with <$50K / ₹40L✅API + WhatsApp + cloud; bootstrappable.

12. Feasibility score

AxisWeightScoreNotes
Problem intensity2017/20Subs go bankrupt over this; “lose money” is literal, not figurative. Just shy of top band because the worst pain clusters at dispute time, not daily.
Demand evidence1512/15Strong: bankruptcy stats, named cases, legal+academic sources. Held under 13 by thin verbatim subcontractor quotes (forums sparse).
Build feasibility1511/15Stack is off-the-shelf; BoQ-aligned extraction accuracy is real, bounded engineering — ~4–5 months.
Distribution clarity1511/15Named lists (Gapensi/Aspeknas), live news targets, lawyer referrers; conversion math still estimated.
Revenue mechanics1511/15Pricing fits wallet; ARR math undemanding. Churn risk: subs may cancel between projects — annual prepay mitigates.
Time to first revenue107/10Paid pilots in 6–10 weeks off targeted WhatsApp outreach; not instant self-serve.
Defensibility106/10Soft moat: accumulating cross-project evidence ledgers, lawyer-referral network, Indonesian-document specificity. Copyable but the incumbents are pointed the other way (owner-side).
Total10075/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

Key assumptions to validate (3–5)

  1. Assumption: Subcontractors will pay ~Rp1.5M/mo for defensive evidence (not productivity). How to test: 30 in-person interviews with Jakarta/Surabaya specialist subs who’ve been burned; pre-sell 10 annual pilots before building.
  2. Assumption: A model can extract BoQ-aligned quantities from a real mandor’s photo + Bahasa voice note accurately enough to be claim-credible. How to test: Collect 200 real site entries; measure extraction accuracy vs. a QS’s manual opname; target ≥85% with easy human correction.
  3. Assumption: Owners/MK reps will tap “acknowledge received” (the asymmetry-breaker). How to test: Run the acknowledgement loop on 5 live projects; measure acknowledgement rate within 72h.
  4. Assumption: The dispute-dossier export materially helps subs recover money (drives word-of-mouth). How to test: Track 10 stuck-termin cases through partner lawyers; measure recovery vs. baseline.

Risk flags

  1. Channel dependency: Heavy WhatsApp Business API reliance — pricing/policy changes hit capture UX. Mitigate with a thin native app fallback.
  2. Adversarial counterparty: Main contractors may pressure subs not to use a “record-keeping” tool against them. Position as neutral progress documentation; lean on owner-side acknowledgement so it’s not framed as anti-GC.
  3. Accuracy liability: A wrong quantity in an evidence packet is worse than no packet. Requires visible confidence, mandatory human confirmation before any packet is “locked.”
  4. Market timing: If infrastructure spend slows, project volume (and willingness to pay) softens with it.

14. Structured verdict

Score:                  75/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder (multimodal/eval discipline) + Indonesian construction-law advisor + a GTM partner with Gapensi/Aspeknas reach
Time to revenue:        8–12 weeks (paid pilots via targeted WhatsApp outreach)
Capital to launch:      ₹25–35 lakh ($30–42K)
Top 3 assumptions to validate first:
  1. Willingness to pay Rp1.5M/mo for defensive evidence — 30 interviews + 10 pre-sold annual pilots
  2. BoQ-aligned extraction accuracy ≥85% on 200 real mandor entries vs. a QS baseline
  3. Owner/MK acknowledgement rate ≥50% within 72h on 5 live projects
Kill criteria:
  - Abandon if <8 of 30 interviewed subs will pre-pay an annual pilot
  - Abandon if extraction accuracy stays <70% after the eval loop, or owner acknowledgement <25% on live projects

15. Next step — 1-week validation sprint

  • Day 1–2: Pull Gapensi/Aspeknas Jakarta + Surabaya specialist-subcon directories; build a 200-firm list; write the Bahasa WhatsApp pitch + 90-sec demo around “termin nggak cair.”
  • Day 3–4: Run 25–30 calls/visits with subs who’ve been burned. Show a fake-door packet mockup. Ask the only question that matters: “Bayar Rp1.5 juta/bulan untuk ini?” — and try to collect 5–10 signed annual-pilot LOIs (deposit if possible).
  • Day 5: Decide go / no-go on a falsifiable threshold: ≥8 of 30 subs commit to a pre-paid annual pilot. Below that, the pain is real but the wallet isn’t — kill or reshape toward the lawyer-referral channel.

Interested in a detailed proposal?

Get a deep-dive with market research, competitive analysis, and implementation roadmap.

Contact us

info@startupbasket.ai