PortClear — pre-shipment rejection screen for Indian agri-exporters

1. One-liner

Catches the label, document, and destination-rule errors that get an Indian agri-export container rejected — before it ships.

2. Trend signal — why now?

Indian agri-food exports hit $51.9B in FY24-25 (rice $12.5B, spices $4.45B, marine $7.2B), growing 6.4% YoY — but the rejection rate at destination is climbing and the cause is overwhelmingly paperwork and labelling, not the food itself.

• UNIDO: labelling errors accounted for 31% of total rejections of Indian agri-food exports in 2024. A separate analysis puts ~57% of recorded rejections as labelling-related (missing allergen declarations, wrong product descriptions, non-compliant nutrition panels, destination-specific formatting).
• EU rejections of non-basmati rice rose 3 → 37 between 2020 and 2024; ~92% of shipments flagged at Rotterdam/Antwerp cite MRL/cert/ETO issues. “Most shipment delays are paperwork-related — not product failures. A mismatched HS code, an inconsistent phytosanitary certificate, or an error in the certificate of origin can all trigger a customs hold.”
• The cost is brutal and per-event: a single ETO/MRL rejection runs ₹50–80 lakh (shipment value + buyer penalties + lost future business). One documented Gujarat curry-powder case: ₹1.2 crore loss — ₹45L shipment recalled across 450 German stores, a 3-year ₹8-crore contract cancelled, manufacturer blacklisted.

The 2026 unlock: multimodal LLMs can now read a label photo + a Certificate of Analysis + the spec sheet, cross-check them against a destination country’s import ruleset, and flag what will get the container held — in minutes, for cents. That capability did not exist cheaply 18 months ago.

Provenance:

• Signal 1 (demand): UNIDO — labelling errors = 31% of Indian agri-food rejections (2024); EU non-basmati rice rejections 3→37 (2020-24); single ETO rejection costs ₹50-80L, Gujarat case ₹1.2cr — https://www.eij.news/post/indian-exporters-face-rejection-of-food-shipments-by-eu-causes-and-solutions-explained-for-exporter , https://sadbhaavspices.com/top-5-spice-export-rejection-reasons/ — 2026-05-29
• Signal 2 (feasibility): Multimodal LLMs read label artwork + CoA + spec vs destination ruleset; today’s MRL screening is manual database subscriptions / expensive lab services (Eurofins, Tentamus), no AI pre-screen exists — https://www.eurofinsus.com/food-testing/resources/mrl-starter-guide-managing-pesticide-maximum-residue-limits-for-global-trade/ — 2026-05-29
• Signal 3 (economic): India agri-export $51.9B FY24-25 (+6.4%); existing export-doc SaaS (Expodite, VisualExportEMS 2500+ clients, EximDoc) proves exporters pay for software but generates docs only — no destination-rule screening — https://www.ibef.org/exports/agriculture-and-food-industry-india , https://www.visualexportems.com/ — 2026-05-29 Category: Geographic arbitrage

3. The opportunity

Two adjacent markets exist and neither solves this:

1. Export-documentation software (Expodite, VisualExportEMS, EximDoc, ExpoMaster — 2,500+ Indian exporters on one vendor alone). These generate the commercial invoice, packing list, and certificate of origin. They are dumb form-fillers: they produce a clean-looking document regardless of whether its contents will get rejected at Rotterdam. They have no model of destination import rules.
2. Lab testing & MRL data services (Eurofins, Tentamus, AGT). These tell you the actual pesticide/aflatoxin/ETO levels in a physical sample. Accurate, necessary, slow, and expensive (₹15-40k per sample, days of turnaround) — and they only test what you ask them to test against the country you remember to specify.

The gap between them is where containers get rejected: the exporter has a document and a lab report, but nobody checks that the label artwork carries the destination’s mandatory declarations, that the declared values clear that specific country’s limits (EU aflatoxin 2 mg/kg vs FSSAI 10 mg/kg; EU ETO <0.05 mg/kg), that the importer details and HS code match, and that every document the destination requires for this commodity is present and consistent. Today that cross-check is a senior export manager’s head, a WhatsApp group, and luck. PortClear is the screen that sits between “docs generated” and “container sealed” and says this will be rejected, here’s why, fix these four things first.

4. Target market

• Primary customer: Owner / export manager at a small-to-mid Indian agri-food exporter — rice mills, spice processors, marine/seafood packers, processed-food and dairy exporters — roughly ₹5–100 crore annual export turnover, APEDA-registered, shipping 5–60 consignments/month, no in-house regulatory-affairs team.
• Why they buy (in their words): “One held container at a US or EU port wipes out a quarter’s margin. We found out the label was wrong after the buyer’s lawyer called.” Rejection isn’t just the ₹45L shipment — it’s the cancelled contract, the blacklist, the insurer raising premiums.
• Rough TAM reasoning: India agri-food export is $51.9B across thousands of exporters; rice alone is 22M tonnes/year. APEDA-registered exporters number in the tens of thousands. Even a serviceable beachhead of ~8,000 small/mid exporters shipping to rejection-prone EU/US/Gulf markets is a multi-thousand-customer SaaS market at SMB pricing.
• Why now for them: EU/US tightened MRL/ETO enforcement sharply in 2024-25 (ETO crisis suspended MDH/Everest blends in Hong Kong/Singapore); RASFF flags India among most-flagged origins; rejection counts are rising YoY. The pain just got materially worse in the last 18 months.

5. Product sketch (MVP)

• Shipment screen: upload the label artwork (photo/PDF), the commercial invoice + packing list, and the CoA/lab report; pick destination country + commodity → get a pass/fail risk report in minutes.
• Destination ruleset engine: per-country, per-commodity checklist — mandatory label declarations, required documents, aflatoxin/MRL/ETO thresholds, importer-detail and HS-code consistency.
• Red-flag report: ranked list of rejection risks (“EU requires lot number + EU importer address on label — missing”; “declared aflatoxin 6 mg/kg exceeds EU 2 mg/kg — do not ship to EU, OK for Gulf”).
• “Test-before-you-ship” prompts: where AI can’t verify a physical-residue claim, it explicitly tells you which lab test to commission for which contaminant against which country limit.
• Label fix suggestions: generates the corrected mandatory-declaration block for the destination, ready to hand to the printer.
• Document-completeness check: flags missing/expired phytosanitary cert, COO mismatch, NPOP organic cert expiry.
• WhatsApp submission: export clerk forwards label photo + CoA to a WhatsApp number, gets the screen back in-thread (matches how Indian export desks actually work).
• Shipment history: every screened consignment archived with its verdict — an audit trail for buyer disputes and insurer queries.

6. AI angle — what’s load-bearing

Remove the AI and there is no product. The core job is reading messy, unstructured inputs and reasoning against rules: a phone photo of a printed label in mixed English/regional text, a lab CoA in a non-standard PDF layout, a spec sheet — extracting declared values and label elements, then cross-checking them against a structured destination ruleset and producing a defensible rejection-risk verdict. That’s multimodal extraction + rule reasoning, exactly the 2026 capability. The ruleset is curated data, but the value is the model turning a sloppy real-world document packet into “this clears EU, this doesn’t, here’s why” without a human regulatory expert reading every page. A dumb form-filler (the incumbents) can’t do it.

7. Localization angle

India-first and it’s the whole wedge. The customer is an Indian export desk that runs on WhatsApp, mixes English with Hindi/Gujarati/Tamil on labels and in messages, prices in rupees, and is APEDA/DGFT/FSSAI-literate but not EU-/FDA-regulation-literate. A generic global “trade compliance” tool priced in dollars and built around US/EU shippers misses this customer entirely. The wedge is being fluent in both ends: the Indian document/cert reality (phytosanitary from NPPO, APEDA RCMC, FSSAI) and the destination ruleset (EU/FDA/Gulf SFDA). WhatsApp-first submission and ₹ pricing are not nice-to-haves — they’re how you reach a 30-person spice processor in Unjha or a rice mill in Karnal.

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

• Pricing: tiered SaaS. Starter ₹2,499/mo ($30, up to 15 screens/mo), Growth ₹6,999/mo ($84, up to 60 screens + label-fix), Pro ₹14,999/mo (~$180, unlimited + multi-user + API for their freight forwarder). Per-screen overage on Starter.
• ACV: blended ~₹70-90k/yr ($850-1,100).
• Rough math to $1M ARR: ~1,000 exporters × ~₹85k ACV ≈ ₹8.5cr ≈ $1.02M. ~3.5% of an ~8,000-exporter serviceable base — not heroic.
• Rough math to $5M ARR: ~4,800 exporters, OR fewer exporters + a per-screen usage tier for high-volume rice/marine packers (60+ consignments/mo) + a freight-forwarder/CHA white-label reselling PortClear to their exporter clients.
• Expansion path: screens/month usage grows with the exporter; add destinations (Gulf SFDA, UK, Japan, Korea) as paid modules; add a “buyer-shareable clearance report” the exporter sends their importer to win trust; eventually a CHA/forwarder API tier.

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

• APEDA / commodity-board exporter directories are public. Scrape the registered rice, spice, and marine exporter lists; segment to small/mid shippers to EU/US. Send a personalized one-pager: “Here are the 3 most common reasons [commodity] from India got rejected at [their top destination] in 2024 — we screen for all of them before you ship.” Expect 3-6% reply at this specificity.
• Spice/rice exporter clusters & associations: Unjha (spices), Karnal/Haryana (basmati), Kochi (marine/spices), Indian Spices Board and IOPEPC/MPEDA member lists. One workshop or WhatsApp-broadcast deal per cluster reaches dozens of owners who all know each other.
• Freight forwarders & CHAs as a channel: they eat the reputational hit when their client’s container is held. Offer them a co-sell / white-label — they already have the exporter relationships and a reason to push a rejection-prevention tool.
• “Rejection post-mortem” content + cold outreach to recently-burned exporters: RASFF and Indian trade-portal rejection alerts name flagged consignments/origins. Reach out to exporters in freshly-flagged categories the week the alert publishes — that’s a hair-on-fire moment.

10. Build complexity — justification

Medium. The screening engine is off-the-shelf multimodal AI for extraction + a reasoning layer over a curated destination ruleset; WhatsApp Business API and a standard web app are commodity. The genuine work is building and maintaining the per-country, per-commodity rulesets (label-declaration rules, MRL/aflatoxin/ETO limits, required-document lists) accurately enough that exporters trust the verdict — that’s domain-expertise-heavy curation, not exotic engineering. A pair (one technical, one with agri-export/regulatory domain depth) ships a credible EU+US, rice+spices v1 in ~3-4 months and expands rulesets from there.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Advisory/screening tool; no regulated authority claimed. Must disclaim it’s not a substitute for lab testing or legal sign-off.
Ethical — no harm / dark patterns	✅	Reduces rejected/wasted shipments; aligns with food safety.
Market exists (evidence above)	✅	$51.9B exports, rising rejection counts, existing doc-SaaS spend.
1–5 person team can build this	✅	Pair + domain advisor, ~3-4 months to v1.
Launchable with <$50K / ₹40L	✅	API + web + WhatsApp + ruleset curation labor. Well under cap.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Per-event loss ₹50L-1.2cr; rising enforcement; recurring per shipment. Genuinely hair-on-fire for a burned exporter.
Demand evidence	15	13/15	Hard third-party stats (UNIDO 31%, RASFF, rice 3→37), documented ₹-cost cases, proven willingness to pay for adjacent doc-SaaS.
Build feasibility	15	11/15	AI/app commodity; ruleset accuracy is the hard, ongoing part — trust depends on it.
Distribution clarity	15	11/15	Named public directories + clusters + forwarder co-sell. Conversion uncertain; needs domain-credible outreach.
Revenue mechanics	15	11/15	₹ pricing fits wallets; ACV reasonable; $1M needs ~1,000 paying exporters — achievable but not trivial to acquire.
Time to first revenue	10	6/10	A burned exporter buys fast, but trust in a verdict tool takes a pilot/proof period; ~6-10 weeks to first paid.
Defensibility	10	5/10	Moat is the ruleset accuracy + accumulated screened-shipment data + cluster trust. Copyable by a funded doc-SaaS bolting on screening — that’s the real threat.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

domain-expertise-required · technical-heavy

Key assumptions to validate (3–5)

1. Assumption: Exporters will trust an AI “this will be rejected” verdict enough to delay/fix a shipment. How to test: Run 20 free retro-screens on exporters’ past rejected shipments — would PortClear have caught it? If it catches ≥70%, that’s the demo that earns trust.
2. Assumption: The screenable failure modes (label/doc/declared-value) are a big enough slice of rejections to matter, given that physical-contaminant rejections need a lab. How to test: Classify 100 documented Indian agri rejections into “PortClear-catchable” vs “lab-only.” Need the catchable slice ≥40% (UNIDO labelling alone is 31%).
3. Assumption: Small/mid exporters (not just big houses) will pay ₹2.5-15k/mo. How to test: 30 in-person interviews across Unjha + Karnal clusters with a live screen demo; target ≥8 pre-orders.
4. Assumption: Ruleset can be curated accurately for EU+US+Gulf, rice+spices+marine without a regulatory team. How to test: Build the EU-spices ruleset, validate against 50 real shipments with a domain advisor; measure false-pass rate (a false pass is the killer).

Risk flags

1. Liability / false-pass risk: If PortClear passes a shipment that then gets rejected, the exporter blames the tool. Must be positioned as a screen that reduces risk, not a guarantee, with hard lab-test prompts for physical contaminants — and the verdict logic must err toward flagging.
2. Incumbent fast-follow: Expodite / VisualExportEMS already have thousands of exporters and the document workflow; they could bolt destination-screening on. The defense is depth + speed + cluster trust before they notice.
3. Ruleset rot: MRL/label rules change constantly (EU updates frequently). Stale rules = wrong verdicts = lost trust. Needs an ongoing curation discipline, not a one-time build — this is the real operating cost.
4. Scope honesty: AI cannot detect actual ETO/aflatoxin in the physical product from a photo. Over-claiming here destroys credibility fast; the product must be crisp that it screens declarations and documents, and routes physical-residue verification to labs.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder + agri-export/regulatory domain advisor
Time to revenue:        6–10 weeks (after a retro-screen proof)
Capital to launch:      ₹6–12 lakh ($7–14K)
Top 3 assumptions to validate first:
  1. Retro-screen ≥70% of exporters' past rejections — would PortClear have caught them?
  2. PortClear-catchable failures ≥40% of documented rejections (label/doc/declared-value vs lab-only).
  3. 30 cluster interviews → ≥8 pre-orders at ₹2.5-15k/mo.
Kill criteria:
  - Abandon if retro-screen catches <50% of past rejections (the screenable slice is too thin to sell).
  - Abandon if an incumbent doc-SaaS ships destination-rule screening at comparable depth before v1.
  - Abandon if <8 of first 50 cluster exporters pre-order after a live demo.

15. Next step — 1-week validation sprint

• Day 1–2: Collect 30–40 real rejected Indian agri shipments (rejection reason + the label/docs where obtainable) from RASFF alerts, trade-portal rejection lists, and 5-6 exporter contacts. Hand-build a minimal EU-spices + EU-rice ruleset.
• Day 3–4: Run a manual/AI-assisted retro-screen on those 30-40 cases. Score: of the rejections, what % would a PortClear screen have flagged in advance? Classify catchable vs lab-only.
• Day 5: Call 15 exporters in the Unjha/Karnal clusters, walk them through 2-3 retro-screens of their own past pain, and ask for a ₹-committed pre-order or LOI.
• Go / no-go: GO only if retro-screen catch rate ≥60% and ≥3 of 15 exporters give a paid pre-order or signed LOI. Falsifiable: a low catch rate or zero pre-orders kills it.