KodiClose — deduction-gap closer for Kenyan SMEs

1. One-liner

Scans your M-Pesa outflows, flags every expense KRA will disallow, and files the reverse-invoice before it costs you tax.

2. Trend signal — why now?

On 1 January 2026 the Kenya Revenue Authority flipped a switch that reprices every informal purchase a Kenyan business makes. From that date, KRA’s iTax system cross-validates every expense line in your income-tax and VAT returns against the eTIMS invoices your suppliers actually transmitted — line by line, automatically, no human reviewer needed. Any expense not backed by an eTIMS invoice is disallowed and taxed as if it were profit. For a business at a 30% effective rate, a KES 100,000 supplier payment with no eTIMS invoice now really costs KES 130,000 (Adamjee Auditors, “eTIMS 2026: KRA Expense Validation Rules”; Sokodirectory, Nov 2025).

This lands on an economy that is largely informal. Over 90% of SME transactions run through M-Pesa, and a huge share of suppliers — the jua kali welders, the Gikomba furniture makers, the mama mboga, the boda transporters — are not eTIMS-registered and have no idea how to be. KRA’s own escape hatch is reverse (buyer-initiated) invoicing: the buyer issues the eTIMS invoice on the supplier’s behalf for suppliers under KES 5M turnover. But doing that manually, purchase by purchase, across dozens of scattered informal suppliers, is a nightmare. One Capital FM report notes some manufacturers have hired up to 100 staff purely to reconcile mismatched invoices against KRA’s records. That’s the pain, quantified.

The Finance Act 2026 also put a floor under enforcement: a minimum KES 100,000 penalty for companies (KES 10,000 for individuals) that fail to use the system. The stick is real and the deadline already passed — this is a live fire, not a “someday” regulation.

Provenance:

• Signal 1 (Demand): SMEs face expenses disallowed / taxed as profit when suppliers aren’t eTIMS-compliant; jua kali suppliers can’t issue invoices; manufacturers hiring up to 100 staff to reconcile — https://capitalfm.africa/etims-threatened-by-system-gaps-supplier-challenges-expert/ — 2026-02
• Signal 2 (Feasibility): KRA offers free reverse/buyer-initiated invoicing API + eTIMS OSCU; M-Pesa Daraja API exposes till/paybill/statement data — cheap multilingual LLM can classify outflows and draft invoices — https://quartet-co.com/reverse-invoicing-etims-kenya/ — 2026 (accessed 2026-07)
• Signal 3 (Economic): SMEs already pay KES 1.5K–5K/mo for eTIMS tooling and KES 10K–30K/yr to accountants; Finance Act 2026 sets KES 100,000 minimum non-compliance penalty — https://adamjeeauditors.com/e-tims-2026-kra-expense-validation-business/ — 2026-01 Category: Regulatory arbitrage

3. The opportunity

The incumbents are full cloud-accounting suites — ZYNO Books, ERPs, the classic bookkeeping tools — that treat “eTIMS compliant” as a checkbox on the sales side (they help you issue invoices to your customers). None of them attack the sharp, painful, brand-new problem: your purchase-side deduction gap. The moment that hurts is filing season, when KRA silently disallows expenses you already paid for because a supplier you’ll never see again didn’t transmit an invoice.

A focused tool does one thing 10× better than the suites: it ingests your actual outflows (M-Pesa statement + bank feed), tells you this specific list of KES-X in payments will be disallowed, and then closes each gap — either by nudging the supplier to issue an eTIMS invoice, or by generating and filing the reverse-invoice for you via KRA’s buyer-initiated flow. It converts a year-end audit ambush into a running, always-current “how much tax am I about to overpay” number. The full suites can’t lead with this because it’s not their wedge and it exposes an ugly truth their onboarding would rather hide.

4. Target market

• Primary customer: Owner-operators and finance clerks at Kenyan SMEs with KES 5M–200M annual turnover — the tier that is VAT/income-tax filing and does buy from informal suppliers: small manufacturers, hardware & building-supply merchants, restaurants and bars, agri-processors, wholesalers, matatu/logistics SACCOs, mid-size retailers. Nairobi, Mombasa, Nakuru, Kisumu.
• Why they buy (in their words): “If purchases have no electronic receipts, KRA assumes additional profit and taxes me on the full amount.” “My suppliers are jua kali — half of them don’t even have a KRA PIN, let alone eTIMS.” They buy because the alternative is paying 30% tax on money they already spent, plus a KES 100K penalty floor.
• Rough TAM reasoning: Kenya has well over a million registered businesses; the filing-obligated SME band (above the KES 5M reverse-invoicing threshold) plausibly numbers in the low hundreds of thousands. Capturing even 3,000–8,000 of them at KES 2K–6K/mo is a KES 70M–580M/yr ($0.5M–$4.5M) business. This is a “big enough to matter, too small and too local for global SaaS to bother” niche — exactly the target profile.
• Why now for them: The rule went live 1 Jan 2026 and bites at the first filing against 2025+ income. This is the first year owners feel the disallowance in their actual tax bill. Urgency is externally scheduled by the state, not manufactured.

5. Product sketch (MVP)

• Connect M-Pesa + bank: Link your M-Pesa till/paybill (Daraja API) and bank statement; KodiClose pulls every outflow automatically.
• Deduction-gap radar: Classifies each outflow as a business expense, matches it against eTIMS invoices KRA actually received, and surfaces a live list: “KES 412,000 of your expenses are currently un-deductible — here’s the KES 123,600 in extra tax that costs you.”
• Supplier compliance check: Tells you, before you pay, whether a supplier is eTIMS-registered — so you stop creating new gaps.
• One-tap reverse-invoice: For under-threshold informal suppliers, drafts the buyer-initiated eTIMS invoice (line items pre-filled from the M-Pesa reference) and files it to KRA. Turns a 20-minute manual chore into one tap.
• Supplier nudge (WhatsApp): For suppliers who should self-invoice, sends a pre-written WhatsApp message with a link to get their invoice transmitted, and chases until it lands.
• Filing-ready pack: At return time, exports a clean reconciliation showing every expense, its eTIMS status, and the closed gaps — hand it to your accountant or upload straight to iTax.
• Penalty-risk score: A single running number: your current disallowance exposure and the estimated tax hit if you filed today.

6. AI angle — what’s load-bearing

The load-bearing AI job is turning a raw, messy M-Pesa/bank outflow log into a structured, eTIMS-ready expense ledger with no manual coding. M-Pesa lines are semi-structured noise — “Pay Bill to KPLC”, “Merchant Payment 8842 to JOHN MWANGI HARDWARE”, “Send Money to 0722…”. A cheap multilingual LLM (English + Swahili + Sheng) classifies each line into expense vs personal, assigns the KRA tax/commodity code, matches it to a supplier identity, and drafts the reverse-invoice line items — the exact work a bookkeeper does by hand and the reason manufacturers are hiring 100 people to do it. Remove the AI and you’re back to a spreadsheet and a clerk. The reconciliation-matching and invoice-drafting is where 90% of the human hours die, and it’s precisely what the model collapses from hours to seconds.

7. Localization angle

This is entirely a localization play — it cannot exist as a generic global product:

• Payment rail: M-Pesa (Daraja API) is the transaction spine; 90%+ of SME payments flow through it. No M-Pesa parsing, no product.
• Regulatory rail: Built directly on KRA’s eTIMS / OSCU and buyer-initiated invoicing APIs and the specific 2026 validation rule. This is deep, jurisdiction-specific knowledge that is worthless outside Kenya — and a moat inside it.
• Language: Supplier names and payment references come in English, Swahili, and Sheng. Supplier nudges go out over WhatsApp, the default business channel.
• Pricing: A KES 2,000/mo tier works where a $49/mo tool would be laughed out of the room.
• Expansion: The same regulatory shape is spreading across Africa — Nigeria’s FIRS mandate pulls SMEs in from Jan 2026, and other markets are following. The Kenya build is the template; the pattern ports even though the specific integration doesn’t.

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

• Pricing: Tiered by turnover / transaction volume. Starter KES 1,500/mo (micro, low outflow count), Growth KES 3,500/mo (the core SME), Pro KES 6,000/mo (multi-till, multi-location, accountant seat). Optional per-reverse-invoice usage fee for very high-volume filers.
• ACV: KES 42,000 ($320) blended per year.
• Rough math to $1M ARR (~KES 130M): ~3,100 customers at the KES 3,500/mo Growth tier. Very achievable inside the filing-obligated SME band.
• Rough math to $5M ARR: ~15,000 customers, or a lower count blended up with Pro tiers + accountant-firm resale + expansion into a second African market (Nigeria/Tanzania/Uganda) running the same disallowance pattern.
• Expansion path: Land on deduction-gap closing → expand into full purchase-ledger + supplier-payment workflows → sell an “accountant console” seat so the SME’s bookkeeper manages 30 clients from one screen (channel + ACV multiplier).

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

• Accountants and tax agents as the wedge (highest leverage). A few hundred small accounting/audit firms serve thousands of these SMEs and are the ones drowning in the reconciliation. Sign 15–20 firms, give them a free multi-client console, and each brings 20–50 clients. This is the fastest path to the first 100.
• KRA-timed WhatsApp + Facebook groups. Kenyan SME owners live in WhatsApp business groups and Facebook groups (“SME Kenya”, “Biashara Kenya”, county trader associations). Post the concrete horror scenario — “your KES 100K supplier payment is now a KES 130K cost” — with a free “how big is your deduction gap?” scan as the hook.
• Free deduction-gap audit. Landing page: upload one M-Pesa statement, get an instant number for how much tax you’re about to overpay. That number is the sales pitch; conversion to paid is the fix.
• Trade-association + SACCO partnerships. Matatu SACCOs, hardware-dealer associations, restaurant associations — one partnership deal reaches hundreds of members who share the identical exposure.
• Deadline-driven urgency ads. Cheap targeted ads around filing dates (VAT monthly, income-tax June) when the pain is acute and searched.

10. Build complexity — justification

Medium. The AI classification and invoice drafting is off-the-shelf (multilingual LLM + prompt/eval work). The real work is the integration surface: M-Pesa Daraja API, bank statement ingestion, and — the gnarly part — KRA’s eTIMS / OSCU and buyer-initiated invoicing APIs, which are jurisdiction-specific, imperfectly documented, and prone to downtime (a known, reported problem). Budget a small team ~3–4 months to a credible v1: the LLM ledger + gap radar can ship early on statement upload alone; live reverse-invoice filing is the harder second milestone. No custom models, no hardware, no research risk.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Buyer-initiated invoicing is KRA’s own sanctioned mechanism; product helps compliance, isn’t regulated itself.
Ethical — no harm / dark patterns	✅	Helps SMEs pay the tax they legitimately owe and no more. No exploitation.
Market exists (evidence above)	✅	Live 2026 rule, penalty floor, manufacturers hiring 100 reconcilers, existing paid tooling.
1–5 person team can build this	✅	Off-the-shelf AI + API integration; ~3–4 months.
Launchable with <$50K / ₹40L	✅	No capex; API + inference costs only.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Hair-on-fire: money already spent gets taxed as profit, KES 100K penalty floor, deadline already live. Buyers overpay 30% on unbacked expenses today.
Demand evidence	15	12/15	Strong: live regulation, quantified pain (100-staff reconciliation), existing paid eTIMS tooling, jua-kali supplier gap widely documented. Docked slightly — no verbatim SME-owner quote yet, mostly expert/press sourcing.
Build feasibility	15	11/15	Off-the-shelf AI; integration-heavy on KRA + M-Pesa APIs, which have documented downtime and thin docs. Not a weekend build.
Distribution clarity	15	12/15	Accountant channel + WhatsApp/SACCO groups + free-audit hook are concrete and cheap. Conversion math still to prove.
Revenue mechanics	15	11/15	Pricing benchmarked to existing eTIMS tooling (KES 1.5–5K/mo); ACV modest, so it needs volume. Path to $1M is clean; $5M needs a second market.
Time to first revenue	10	7/10	Deadline urgency + free-audit funnel enables fast paid conversion, but the reverse-invoice filing piece takes a couple months to be trustworthy.
Defensibility	10	5/10	Moat is regulatory/integration know-how + accountant lock-in + accumulating supplier-compliance data. Copyable by a local suite eventually; head start and focus win the niche.
Total	100	75/100

13. Qualitative modifiers

Founder-fit tags

domain-expertise-required · technical-heavy — you need someone who genuinely understands KRA eTIMS mechanics and Kenyan SME accounting, paired with an engineer comfortable wrangling M-Pesa/KRA APIs. A Nairobi-based founder or a diaspora founder with a strong local co-founder is the ideal shape.

Key assumptions to validate (3–5)

1. Assumption: SMEs will pay KES 3,500/mo to close their deduction gap. How to test: Run the free-audit funnel on 50 SMEs, show each their disallowance number, and pre-sell a paid fix. Target ≥20% intent-to-pay.
2. Assumption: KRA’s buyer-initiated invoicing API is stable and accessible enough to file reverse-invoices programmatically at volume. How to test: Get sandbox/production access, file 50 real reverse-invoices, measure success rate and downtime.
3. Assumption: Accountants will resell/refer rather than see this as a threat to their fees. How to test: Pitch 15 small firms the free multi-client console; measure how many onboard ≥10 clients in 30 days.
4. Assumption: M-Pesa Daraja data is rich enough to classify most outflows to a supplier identity. How to test: Parse 20 real SME statements; measure % of outflows auto-classified with correct tax code without human touch.

Risk flags

1. Platform dependency: Deep reliance on two APIs you don’t control — KRA eTIMS/OSCU (documented downtime) and M-Pesa Daraja. An API change or outage is an existential operational risk; needs graceful-degradation design (upload fallback).
2. Regulatory whiplash: KRA could shift thresholds, extend grace periods, or change the reverse-invoicing flow, reshaping the pain overnight. Being close to the regulator’s roadmap is essential.
3. Incumbent fast-follow: A local accounting suite (ZYNO, ERP vendors) could bolt a “deduction-gap” view onto their existing distribution. Speed and a sharper, purchase-side-only wedge are the only defense in the first year.

14. Structured verdict

Score:                  75/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Nairobi-based (or diaspora + local co-founder) domain expert on KRA eTIMS + one engineer fluent in M-Pesa/KRA APIs
Time to revenue:        8–14 weeks (free-audit funnel → paid; reverse-invoice filing follows)
Capital to launch:      KES 1.5M–3M ($10K–$22K)
Top 3 assumptions to validate first:
  1. SME willingness-to-pay KES 3,500/mo — pre-sell against a shown disallowance number, target ≥20% intent
  2. KRA buyer-initiated invoicing API stability at volume — file 50 real reverse-invoices, measure success/downtime
  3. Accountant channel adoption — onboard 15 firms to a free console, measure clients brought in 30 days
Kill criteria:
  - Abandon if <15% of 50 SMEs shown their deduction gap express intent to pay
  - Abandon if the KRA reverse-invoicing API can't be filed against reliably (>20% failure) with no viable workaround
  - Abandon if a well-distributed local suite ships an equivalent purchase-side deduction-gap tool before your v1

15. Next step — 1-week validation sprint

• Day 1–2: Build a dead-simple “deduction-gap audit” — upload one M-Pesa statement (or paste it), classify outflows with an LLM, cross-check a sample against eTIMS-registered supplier lookups, and output a single number: “KES X of your expenses are currently un-deductible, costing you ~KES Y in extra tax.” No accounts, no polish.
• Day 3–4: Get it in front of 40–50 real Kenyan SME owners and 10 accountants via WhatsApp SME groups and direct outreach. Each runs their statement. Capture the disallowance number and ask: “Would you pay KES 3,500/mo for a tool that closes this automatically?”
• Day 5: Decide. Go if ≥20% of SMEs shown a non-trivial gap say they’d pay, and ≥5 accountants ask to put their clients on it. No-go if the gaps are small (rule is toothless in practice) or intent-to-pay is under 15%.

Falsifiable result: a hard intent-to-pay percentage against a shown, personalized tax-loss number — not “people liked it.”