DRAFT — pre-evidence-sprint — not for external distribution
Seed · Bangladesh · Retail Shop OS

The whole shop, from the phone he already owns.

AyByay is the Bangla-first, mobile-only operating system for Bangladesh's paper-run shops — billing, stock and the baki ledger on the phone in the owner's pocket — and AI does the data entry that has killed every prior attempt to put these shops on software. We are pre-revenue by design, raising a milestone-gated seed to answer the only two questions worth money: will a shop pay, and will it stay?

Solo-built, live in production Android in Google Play closed testing Founder runs a real cloth shop on it daily
7
shops signed up
2 in genuine daily use today
0
paying — by design
willingness-to-pay is what the raise proves
1,626SKU
imported in one session
a computerized shop, billing the same day
$150–300K
milestone-gated seed
18–20 months of runway
The wall every attempt has died on

Millions of shops still run on paper — and nobody wants to type in their inventory.

A torn khata for credit, stock in the owner's head, profit a feeling, an hour of midnight reconciliation after fourteen hours on his feet. The software to fix this has existed for years. The reason it never stuck is not price or features — it is data entry: typing a shop's catalog in by hand, or paying someone ৳10–20k to do it. AI is the first tool that can delete that wall — and 2024's collapse in inference cost is the moment it became affordable.

We delete data entry from two directions — one proven live, one being proven on camera this quarter: a computerized shop's whole catalog migrates in one AI-mapped session; a paper shop builds his by photographing purchase memos and scanning barcodes against a shared catalog. We headline what is proven and flag what is promised.

Traction — every number survives a live database check

Seven real shops. An honest funnel, not a headline count.

One definition throughout — active = at least one sale in the trailing 7 days. The last rung is the whole reason to fund the round.

Two internal tenants (demo, test) excluded. The only rows in the payments table are two SSLCommerz sandbox ৳299 test charges on an empty shop — integration testing, not revenue. We show you that so you don't wonder what else was hidden.

Daily-use proof jamaltara — a real cloth shop

Built by hand at the counter, day after day.

550
products
58
sales
32,813
Taka through the till

Active 17 of 30 days in July — front-loaded, honest gaps. "17 of 30" out-credibles "daily."

Baki, precisely: 3 customers carrying ৳585 in active credit (of 21 customers on file). One stored SQL definition, reproducible live.

Catalog-import proof banbro — a computerized grocery

A PrimeERP export, AI-mapped in one session — the migration proof. Not a daily-use shop.

banbro is the one-session migration proof (DokanePOS's turf, not our wedge): 15 sales over 7 active days, then quiet. We describe it as onboarding proof and nothing more.

Market & model

A bounded shop market, a ladder priced for it — the round buys proof, not scale.

Line-of-sight, not a fake bottom-up

Reconciling the macro number with a 400–600-shop seed target.

Establishments, all sectors context only BBS 202411.7M
Retail & wholesale shops with a smartphone-owning operator (~73% of mobile users)Millions
Series-A-scale ARR at 1–2% penetration × ~৳400 blended ARPU1–2%
Seed target <0.1% proof-sliver — the two unknowns, cheaply400–600

The ladder — the shopkeeper only ever buys the first rung

Feature breadth is the monetization ladder. Economics assume the list schedule; Basic runs at a ৳99 launch price today — a step we disclose, not hide.

Land0Free · 100-product cap
Pay99Basic · list ৳299
Pro699Full shop OS
Expand1,499Multi · up to 3 branches

14-day free trial, no card · pay via bKash / Nagad. Switching cost accrues as the shop's whole history lives here.

Why we win — execution, not being first

The ledger layer is won. The full shop OS is still unwon.

The category is fragmented, not empty. TallyKhata won the ledger — but it deliberately stops at the khata. The full shop OS — POS + stock + baki + AI onboarding + offline, in Bangla, on the shopkeeper's phone — is still up for grabs, split between 1998-era desktop POS and sub-scale mobile challengers. AyByay is built to win that layer.

Job to be done AyByay TallyKhata Desktop POS banijjo Hishabee
Whole shop from the phone
Baki ledger in Bangla~
Barcode POS + live stock
AI does the data entry~
Works offline through load-shedding~
5M-listing BD catalog head-start

TallyKhata reports ~9.7M downloads / 5.5M businesses — proof of BD adoption, and proof it won't build the counter itself. banijjo is a real, contemporaneous same-category rival (an old 2012 domain repurposed into a 2025–26 SaaS), pre-traction (~100+ app installs) — not an earlier winner. Our edge over it is concrete: offline-first (banijjo advertises none), AI product-onboarding + the 5M-listing catalog head-start, and a fuller shipped stack. The 5M catalog is a public+licensed head start (~84% public scrape) — not itself a moat; the moat is the correction loop live counters write on top.

No khata or shop-OS company in South Asia has yet returned venture capital. India proved shopkeepers adopt this explosively (Khatabook, OkCredit) and proved free-love doesn't convert. Vyapar is the closest survivor — private, roughly profitable on about a third of Khatabook's capital, by charging for the whole shop OS from day one. That is the model we run — in Bangla, on mobile, with an AI onboarding weapon the Indian cohort never had.

The ask — buying the answer to two questions

$150–300K, milestone-gated. We show the arithmetic against ourselves.

$150–300K
raise · 18–20 months · milestone-gated
400–600
paying shops at the milestone
~৳120–180K
milestone MRR (covers ~6–20% of burn — a Series-A need remains)

CAC per paying shop — the honest band

From activation 20–35% × free→paid 10–20% = 2–7% net conversion at ৳300–600 a visit — a ৳4,286 floor. We publish the real band, not a number our own funnel refutes.

Defensible stretch target: ≤ ৳5,000, contingent on pushing net conversion toward the 7% ceiling.

LTV : CAC — underwater today, on purpose visible

Net LTV ৳2,360–4,720 (net of ~21% commission — never the gross band under a "net" label) against that CAC.

1.5× target

Ratio ~0.1–1.1× — below break-even in all but the best corner. Moving it above 1.5× (annual prepay lifting ARPU, hardware margin, conversion ≥5%) is exactly what the seed buys. This is the honest reason to fund the round, not a red flag we hide.

Use of funds

40% agent network
25% first hires (continuity)
20% founder + ops
15% infra / AI / SMS

Agent-network line separates a costed hardware attach and 1–2 field-lead salaries from the ~৳500-per-prepay commission pool — the channel can't manufacture fake CAC. Retention is measured at 4/8/12 weeks, not assumed at month six; the structure is the standard offshore holdco + BD operating subsidiary.

Hit the milestones and Series A funds a machine with known unit economics.

Miss them and we spent very little finding out. That is the whole bet — a founder who did the math against himself, in Bangla, on mobile, with the one weapon the graveyard never had.

DRAFT — pre-evidence-sprint — not for external distribution. Every figure here is reproducible on demand from the live tenant database (evidence.sql) or a cited public source.