SB StartupBasket
All ideas
66 /100 VALIDATE Medium complexity

GiữMã — code custodian for Vietnam durian cooperatives

Keeps each durian export code's households, authorisations and volumes inspection-ready, and flags the lot that would lose it.

AgriTech SaaS SEA Zalo-firstCompliance-drivenSMBExportAI-agentVietnamese-languageCooperative-channel
— views
Evaluation Scores
66/100

VALIDATE

Overall Score

16
Problem
11
Demand
10
Build
9
Distrib.
8
Revenue
6
Time
6
Defense

GiữMã — code custodian for Vietnam durian cooperatives

1. One-liner

Keeps each durian export code’s households, authorisations and volumes inspection-ready, and flags the lot that would lose it.

2. Trend signal — why now?

Vietnam’s durian trade is the biggest fruit export line the country has ever had, and the thing that gates every kilo of it is a string of characters called a growing-area code (mã số vùng trồng) plus a packing-facility code (mã số cơ sở đóng gói). Nine dated things happened to those codes in the last four months.

  • Exports are at a record. US$4.7 billion in the first seven months of 2026, up 23% year on year, with July the first month over US$1 billion. Durian is over half of all fruit and vegetable exports (The Star, 31 Jul 2026).
  • China is pulling codes. From 12 June 2026 China suspended 15 packing plants and 18 orchards after 77 cadmium-exceedance shipments across 33 packers and 40 orchards; cumulative revocations stand around 55 growing-area and 61 packing codes (Produce Report, June 2026). Domestically, over 400 durian codes drew warnings and 167 were suspended between January 2025 and May 2026 (Tiền Phong, 1 Aug 2026).
  • The regime flipped to self-declaration on 31 July 2026. Resolution 36/2026/NQ-CP moved code issuance to the commune chairman, put the whole procedure online, dropped the 10-hectare minimum, and made the holder operate on “tự khai, tự chịu trách nhiệm” (self-declare, self-answer). The commune must post-inspect within three months of issue and can suspend, restore or revoke (Dân Việt, 31 Jul 2026).
  • Penalties now exist. Decree 214/2026/NĐ-CP sets fines of ₫500,000–20 million for code-management violations and treats “sử dụng mã số vùng trồng hoặc mã số cơ sở đóng gói không đúng nguồn gốc hàng hóa để phục vụ xuất khẩu” as a serious act with mandatory destruction of the goods (Nông nghiệp Hữu cơ, 2026).
  • The police are in it. On 6 August 2026 Deputy PM Hồ Quốc Dũng ordered the Ministry of Public Security to prosecute the buying, renting, forging and fraudulent use of codes, required the Agriculture Ministry to report weekly to the PM, and set a September 2026 deadline for the amended Decree 38 (Dân Việt, Aug 2026). This followed the 28 July arrest of a deputy minister over fraudulent export certificates.
  • Holders are handing codes back. On 31 July 2026 Lâm Đồng’s agriculture department reported four companies voluntarily returning 21 codes; one because “không còn đủ năng lực tài chính nên chủ động xin hủy mã số để phòng ngừa các rủi ro pháp lý phát sinh” (no longer able to fund the obligations, so cancelling the codes to avoid legal risk), another because it was re-signing every farmer linkage contract “theo quy định mới” and re-verifying its household list (Tiền Phong, 31 Jul 2026).
  • The code is being stolen from under the farmers. VOV’s two-part investigation (15 Aug 2026) found a Krông Pắc cooperative that spent ₫1.5 billion registering 26 codes between 2020 and 2026 while brokers exported other people’s fruit under them, and traders deducting ₫2,000–3,000/kg from farmers “để chi cho mã vùng” (to pay for the code).
  • The compliant orchard earns a premium. Exporters facing the cadmium gate are paying ₫5,000–7,000/kg above market only to orchards whose test results pass, ₫43,000–45,000 against ₫30,000–35,000 (Dân Việt, 9 May 2026). A coded orchard alone is worth “vài nghìn mỗi kilôgam” more (VOV, 15 Aug 2026).
  • Electronic traceability went national on 1 July 2026 with durian as the pilot crop. As of 30 August 2026, 1,599 growing-area codes and 281 packing codes were operating on it (VietnamPlus, 3 Sep 2026).

Provenance:

3. The opportunity

Everyone is selling something adjacent and nobody is selling the job.

  • Registration consultancies (Q-CERT, CLV, UCC and a dozen provincial outfits) sell the one-time file: technical declaration, farmer roster, VietGAP paperwork. Quote-based pricing, no ongoing service. Once the code is issued they are gone, and under Resolution 36 the file is now a three-day online self-declaration anyway. Their product just got smaller.
  • Farm-diary apps (FaceFarm, NextFarm, the Farmers’ Association’s ₫2.5B pilot) sell the household a place to log sprays. That is one input. They do not know which of the 100 households on a code are the ones about to fail a cadmium test, who has an authorisation to ship under the code this week, or how many tonnes have already been declared against it.
  • VNPT’s growing-area solution is sold to provinces: officials view files, approve, suspend, run statistics. Its enterprise-facing VNPT Green app submits registration files and “tự động báo cáo định kỳ cho khối chính quyền” (auto-files periodic reports to the authorities). It reports up; it does not defend the holder.
  • The national traceability system (traceviet.mae.gov.vn, live since 1 July 2026) is a consumer QR. The Ministry’s own launch coverage says it “does not prevent production or sale”; it only shows origin. It is where your data ends up, not where your risk is caught.
  • The Đắk Lắk Durian Association’s database project is an association initiative with no product, price or alert layer described.

The gap is the holder’s exposure, and it has four moving parts that no incumbent tracks together:

  1. Households. A code covering 60 hectares and 100 households (VTV’s sample cooperative) is one asset with 100 hands on it. The farmers’ own worry, verbatim: “đến khi cơ quan chức năng đi hậu kiểm phát hiện vi phạm rồi thu hồi mã thì thiệt hại còn lớn hơn gấp bội” (when post-inspection finds a violation and revokes the code, the damage is many times bigger). The holder needs to know which household’s diary is empty, which one bought a phosphate fertiliser with a cadmium warning, which one sprayed a banned active, before the commune officer does.
  2. Authorisations. The rule described by the Đồng Tháp Durian Association: a cooperative delegates its code to one packing-coded enterprise, not to third parties, and the delegation must be returned when the enterprise stops buying. Forged authorisation letters and seals are exactly how Lâm Đồng’s 12 codes got revoked on 28 July 2026. There is no register of who holds a live authorisation on a given code, for how much, until when.
  3. Volume. Ten hectares yields 200–250 tonnes a season. When 600 tonnes ship under a 10-hectare code, that is code laundering, and it is now what the Ministry says it will check. Enterprises are told to “ghi rõ số lượng thu hoạch và số lượng xuất khẩu” (record harvest quantity and export quantity) per orchard. Nobody gives the holder a running meter.
  4. Status. Codes get warned and suspended by GACC and by PPD, published as articles and Excel files on ppd.gov.vn. The Krông Pắc farmers found out their code was being used by others from a journalist. The holder should find out from a Zalo message the morning the list changes.

GiữMã owns all four. It is the file the commune chairman asks for at the three-month post-inspection, the evidence the holder needs when the code is wrongly suspended and must be restored, and the early warning that stops the one household from costing the other 99 their ₫5,000/kg.

4. Target market

  • Primary customer: The director of a durian cooperative or the compliance lead of a linkage company (công ty liên kết) in Đắk Lắk, Lâm Đồng, Gia Lai, Đồng Nai, Tiền Giang or Đồng Tháp holding 1–30 growing-area codes across 20–100 households, trading 300–5,000 tonnes a season. Secondary: the 281 packing houses operating on the electronic traceability system, which receive fruit from many codes and carry the packing-code exposure.
  • Why they buy: In their words. “Mã số vùng trồng giống như tờ giấy chứng minh thư vậy. Thế nhưng người nông dân đâu có biết mà được sử dụng nó đâu, vì bị lấy mất mã số rồi” (the code is like an ID card, but the farmers never got to use it, because it was taken from them). “Nhiều vườn làm ăn uy tín, chăm chút cẩn thận nhưng lại bị kẻ gian mạo danh mã số để gom hàng trôi nổi kém chất lượng mang đi xuất khẩu” (honest orchards get their code impersonated to export floating, low-grade fruit). “Từ mã số vùng trồng ở đây, doanh nghiệp hay các nơi người ta đưa hàng sầu riêng tùm lum về xuất khẩu” (companies bring in durian from all over and export it under our code). The cooperative founder in Krông Pắc: “Họ buôn bán từ năm 2022 đến giờ chiếm lợi hàng trăm tỷ đồng, làm thiệt hại cho hàng nghìn hộ nông dân” (they have traded codes since 2022, taken hundreds of billions of đồng, and harmed thousands of households).
  • Rough TAM reasoning: 1,599 durian growing-area codes and 281 packing codes on the national system as of 30 Aug 2026, with China approving 567 more on 14 Aug 2026 and provinces racing to code 100% of area by year end (Đắk Lắk is at 16.5% of 45,000 ha). Holders average several codes each (the Krông Pắc cooperative holds 26), so call it 600–900 durian holders today, heading to 1,500+ as Resolution 36 removes the 10 ha floor. Across all crops the country has 9,546 growing-area and 1,525 packing codes, which is the year-two market.
  • Why now for them: The 10 ha minimum forced households into shared codes; Resolution 36 keeps those shared codes alive but makes the holder personally answerable, gives the commune three months to come and look, and puts Decree 214 fines and a police directive behind it. The Tây Nguyên harvest runs August to October. The inspection wave lands in Q4 2026.

5. Product sketch (MVP)

  • Code register. Every code the holder owns or uses: type, crop, market, commune, hectares, households, current status pulled from the PPD-published lists, and the yield ceiling implied by the area.
  • Household file on Zalo. Each linked household gets a Zalo thread. Photograph the linkage contract, the land certificate, the paper diary page, the fertiliser or pesticide receipt; the product reads them, files them against the household and the code, and asks for what is missing.
  • Diary risk flags. Each entry is checked against the banned-active list and the pre-harvest interval; phosphate fertilisers and known cadmium-risk inputs are flagged; a household with no entries in 30 days during the season is flagged as an empty diary, which is what inspectors and Chinese officials both look for.
  • Authorisation ledger. Who may ship under this code, for what tonnage, until what date, with the signed letter attached; one-delegate rule enforced; an alert when a delegate’s packing code is warned or when the delegation should have been returned.
  • Harvest meter. Tonnes harvested per household and tonnes declared for export per code, against the area-implied ceiling, so the holder sees “code X at 92% of plausible yield” before customs does.
  • Status watch. Daily check of PPD’s approval, warning and suspension publications for every code in the register; Zalo alert the morning a code, a delegate’s code or a lab on the holder’s shipments changes status.
  • Post-inspection packet. One button produces the commune’s file: household list with contracts, diary coverage, input records, authorisations, harvest and export totals, test certificates per lot.
  • Restoration draft. When a code is suspended, a draft of the corrective-action file the holder submits to get it restored, prefilled from the register.

6. AI angle — what’s load-bearing

The inputs are handwritten diary pages, photographed contracts, receipts with pesticide trade names, and land certificates, all in Vietnamese, arriving over Zalo from 100 households who will not fill in forms. Reading those into structured records is the whole product; a form-based app is what the farm-diary vendors already sell and what the households already ignore.

The second load-bearing piece is judgement: mapping a trade name on a receipt to an active ingredient and to the banned or restricted list for the China market, applying the pre-harvest interval to the planned harvest date, and reading the PPD’s warning articles and Excel files to decide whether this code or this delegate is affected. Remove the model and you are left with a spreadsheet the cooperative director already has and does not keep.

7. Localization angle (if any)

This is a Vietnam-only product by construction.

  • Language and script: Vietnamese throughout, including the reading of handwritten diaries. Product surfaces in Zalo, which is where the cooperative’s household group already lives.
  • Rails: Pricing in đồng, bank transfer or a QR to the cooperative’s account. No card required.
  • Local pricing: ₫1–2 million a month for a cooperative with up to 10 codes is inside what they already lose on a single lot’s ₫2,000/kg “code fee” deduction.
  • Regulatory quirks: The product is built on Resolution 36, Decree 38 (as amended in September 2026), Decree 214, and the Ministry’s own list formats. That is the moat and the maintenance burden.
  • Distribution: Provincial durian associations (Đắk Lắk, Đồng Tháp), commune agricultural officers, and the Farmers’ Association pilots. Not app stores.

Second geography, if it works: Thailand’s durian growers face the same GACC codes and cadmium gate. Later, not now.

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

  • Pricing: Cooperative tier ₫1.5 million/month (≈US$60) for up to 10 codes and 150 households. Company tier ₫4 million/month (≈US$160) for unlimited codes, multiple delegates and the harvest-export meter. Packing-house tier ₫6 million/month (≈US$240) with per-lot test-certificate tracking and multi-code intake.
  • ACV: Blended ≈US$1,400 a year across a 60/25/15 mix.
  • Rough math to $1M ARR: 700 paying holders and packers at the blended ACV. That is roughly half of today’s durian holder universe plus a fifth of packing houses, which is why this is a VALIDATE and not a GO on durian alone.
  • Rough math to $5M ARR: Requires the second crop wave. The same code regime covers mango, longan, jackfruit, passion fruit, banana, coconut and dragon fruit: 9,546 growing-area and 1,525 packing codes nationally. At 2,500 paying holders and a company-heavy mix (blended ACV US$2,000) the number is reachable; the alternative is a Thailand port.
  • Expansion path: More codes per holder as Resolution 36 removes the 10 ha floor and Đắk Lắk codes the remaining 83% of its area; per-lot test-certificate and cadmium-result tracking as an add-on for exporters; a buyer-facing “verified code” view that a Chánh Thu or an NNT Import-Export pays for to source only inspection-ready codes.

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

  • The Excel is the lead list. PPD publishes GACC-approved codes as downloadable spreadsheets with holder names and communes (the 24 Jan 2024 file alone had 708 growing areas and 168 packing facilities; the 14 Aug 2026 approval added 401 and 166). Pull every durian holder in Đắk Lắk, Lâm Đồng and Gia Lai, cross with the warning and suspension notices, and call the ones whose codes were warned in the last 18 months first. Expect 30–40% to take a Zalo call in harvest season; close 10% of those on a free post-inspection packet.
  • Provincial associations. The Đắk Lắk Durian Association (chair Lê Anh Trung publicly calling for “faster digitalisation”) and the Đồng Tháp Durian Association (chair Võ Tấn Lợi, source of the “cha chung không ai khóc” line) run member Zalo groups and just hosted 63 Chinese buyers at the August festival. Offer both a member rate and a co-branded “inspection-ready” badge. One association endorsement is 30–50 cooperatives in a week.
  • Commune officers. Resolution 36 handed the post-inspection to commune chairmen who, per the traders’ petition in Tuổi Trẻ, lack “kỹ thuật và hạ tầng CNTT”. Give every commune agricultural officer in the top 20 durian communes a free reviewer login that reads the holder’s packet. The officer then tells the holders what to bring.
  • The ones who handed codes back. Four Lâm Đồng companies returned 21 codes on 31 July 2026 citing legal risk and the cost of re-verifying household lists. They are the sharpest expression of the pain and are named in print. Call them with the household-file demo.
  • Exporters who pay the premium. NNT Import-Export and Chánh Thu are on record paying ₫5,000–7,000/kg only to orchards that pass. Offer them the buyer view free for six months in exchange for pushing GiữMã to their linked cooperatives.

10. Build complexity — justification

Medium. A Zalo Official Account plus mini app, a multimodal model reading Vietnamese handwriting and receipts, a code and household register, an authorisation and harvest ledger, and a daily scraper of PPD’s published lists and articles. The gnarly parts are the banned-active mapping (trade name to ingredient to market rule) and keeping the packet format aligned with the September 2026 Decree 38 amendment. A pair with one Vietnamese-speaking founder ships a first packet in 10–12 weeks; the status watch and harvest meter land in the following month.

11. Gating checklist

GatePass?Note
Legal in target market✅Helps holders meet Decree 38/Resolution 36 duties; no regulated activity performed
Ethical — no harm / dark patterns✅Protects the households whose codes are being stolen; cannot be used to launder codes because the volume meter is the point
Market exists (evidence above)✅400+ warnings, 167 suspensions, codes returned, ₫1.5B spent by one cooperative, ₫5–7k/kg premium
1–5 person team can build this✅Zalo + model + register + scraper
Launchable with <$50K / ₹40L✅Two founders, one motorbike, one harvest season

12. Feasibility score

AxisWeightScoreNotes
Problem intensity2016/20Losing a code loses the premium on hundreds of tonnes and now carries fines and a police file; but the pain is seasonal and lands on the holder via third parties, so it is felt hard three months a year rather than daily
Demand evidence1511/15Press saturation, dated government numbers, holders returning codes, an underground market priced per kg; no one has yet said “I would pay for software”, and the willingness-to-pay anchor is consultancy spend, not SaaS
Build feasibility1510/15Standard stack, but Vietnamese handwriting, pesticide mapping and a moving decree; 10–12 weeks to a credible packet
Distribution clarity159/15Named lists, named associations, named communes; conversion in Tây Nguyên will need a person on a motorbike, not a landing page
Revenue mechanics158/15₫1.5–6M/month is affordable, but $1M ARR needs half the durian holder universe; the multi-crop expansion is the real path and is unproven
Time to first revenue106/10Harvest is now and associations move fast, but cooperatives pay slowly; 8–12 weeks to the first paid packet
Defensibility106/10Accumulated per-code household and volume history is hard to rebuild; association relationships compound; VNPT or the Ministry could bundle a holder view and cap the market
Total10066/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · sales-heavy · domain-expertise-required

A Vietnamese-speaking founder who can sit with a cooperative director in Krông Pắc, plus an engineer who can make a model read a wet diary page. Without the first, this does not sell.

Key assumptions to validate (3–5)

  1. Assumption: Cooperative directors, not only companies, will pay ₫1.5M/month. How to test: 20 Zalo calls to holders from the PPD list in Đắk Lắk and Lâm Đồng; offer the packet at ₫1.5M for the season; count deposits, not nods.
  2. Assumption: The commune post-inspection under Resolution 36 actually happens within three months and asks for a file. How to test: Ask five commune agricultural officers what they check and what they have been given to check with; obtain one real inspection record.
  3. Assumption: Households will photograph diaries and receipts into Zalo rather than ignore it. How to test: Run one cooperative of 30–60 households for four weeks; measure the share of households with at least one entry a week.
  4. Assumption: PPD’s warning and suspension publications are complete and timely enough to power a status watch. How to test: Reconcile three months of notices against the 12 June 2026 GACC suspension list and the Lâm Đồng 28 July revocations; measure lag.
  5. Assumption: The amended Decree 38 (due September 2026) keeps holder-side record duties rather than moving everything into the national system. How to test: Read the amendment the week it lands; if the state builds the holder file, this becomes a features race against VNPT.

Risk flags

  1. Platform dependency: The Ministry’s traceability system and VNPT Green could ship a holder-side view with alerts. Their incentives point at the state, not the holder, but the Ministry is reporting weekly to the PM and has budget.
  2. Regulatory risk: The September 2026 Decree 38 amendment could change record formats, or the government could sanction the trade harder and shrink the shared-code population that needs this most.
  3. Market timing: Cadmium is a soil problem that “không thể làm 1–2 tháng là xong” (cannot be fixed in one or two months, Võ Tấn Lợi). If China tightens further and volumes fall, holders cut discretionary spend before they cut labour.
  4. Collection risk: Vietnamese cooperatives pay late and seasonally. Price per season, collect up front.

14. Structured verdict

Score:                  66/100
Verdict:                VALIDATE
Confidence:             Medium
Best-fit builder:       Vietnamese-speaking founder with agri-export or cooperative experience, plus one engineer comfortable with multimodal extraction and scrapers
Time to revenue:        8–12 weeks (harvest season, association channel)
Capital to launch:      ₫400–700 million (US$16–28K): two founders for four months, motorbike, model spend, Zalo OA fees
Top 3 assumptions to validate first:
  1. Cooperatives pay, not only companies — 20 calls from the PPD list, count season deposits
  2. Commune post-inspections happen and ask for a file — 5 officer interviews, one real record
  3. Households will feed Zalo — one cooperative, four weeks, ≥50% weekly participation
Kill criteria:
  - Abandon if fewer than 3 of 20 holders will pay ₫1.5M for the season packet by week 4
  - Abandon if the September 2026 Decree 38 amendment moves the household, authorisation and volume records into the national system with a holder login and alerts
  - Abandon if fewer than 40% of households in the pilot cooperative post a diary entry in any given week of the harvest

15. Next step — 1-week validation sprint

  • Day 1–2: Download the PPD durian code spreadsheets for Đắk Lắk, Lâm Đồng and Gia Lai; join the two association Zalo groups; build a list of 60 holders, flagging the ones whose codes appear in any 2025–2026 warning notice and the four Lâm Đồng companies that returned codes. Call 20. Ask three questions: who holds your authorisations, how many tonnes have you declared against each code this season, and what will you hand the commune officer at the three-month inspection.
  • Day 3–4: Take the three most exposed holders and build their packet by hand from photographs sent over Zalo: household list, contracts, diary coverage, inputs, authorisations, harvest totals. Time it. Show it to one commune agricultural officer and ask whether it is what they would check.
  • Day 5: Go / no-go on two numbers: at least 3 of 20 holders agree to pay ₫1.5M for the season packet with a deposit, and at least one commune officer confirms the packet matches what the post-inspection looks for. Fewer than that and the code holders are not the buyer; go back and test the packing houses instead.

Interested in a detailed proposal?

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

Contact us

info@startupbasket.ai