What should open to the public first.
Product research from 5 Nov 2026. One recommendation, the evidence behind it, and the reference data. Written for Stanley. Published openly.
The answer
- Open first
kTown merchant QR
Sats in, M-Pesa out. Never holds money. A Kisumu pilot of 10 shops, free, four weeks.
- Not yet
Ye₿ente wallet
The wallet works, but it holds users’ sats. Opening it wider waits for legal advice under Kenya’s VASP Act.
- Stay inside
Everything else
Marketing engine and ops tooling stay internal. Learning content stays free. Donations stay open.
This is research, not legal advice. Points marked to confirm need a check before anyone acts on them.
Top risks
- !Kenya now licenses virtual asset services.
Every service here gets legal advice before it opens to more people.
- !One Mac over a tunnel is a single point of failure.
If the Mac sleeps, the node goes dark. Fine for a pilot. Not for strangers at scale.
- !Words must stay honest.
Ye₿ente is custodial: Altradits holds the sats and can freeze sends for safety. The homepage wording was fixed in Nov 2026. No page may claim otherwise.
- !Tando is partner and rival.
kTown rides Tando's rails. Tando could build merchant QR itself. The edge is local trust and Kiswahili support, not tech.
- !One person caps everything.
Support, liquidity, and incidents all land on Stanley. Caps on users and amounts keep one bad day survivable.
What exists today
Read from the code on 5 Nov 2026, not from guesses. Live state checked the same day.
Capability inventory
| Capability | State | Depends on | Verdict |
| Ye₿ente wallet: PIN, handles, Lightning address, recovery | Live | Mac node, email | Friends and family only |
| Send Mobile to M-Pesa via Tando | Live | Tando, node liquidity | Inside the wallet only |
| Buy sats: Daraja, Bitika, Pesapal | Built, sandbox, hidden live | Partner approvals | Not for strangers |
| kTown merchant QR | Built, off live | Tando API only | Pilot next |
| Lightning address proxy | Live | Node | Part of the wallet |
| Donate page and Tip API | Live | Node, WoS fallback | Open, keep |
| Paid contact form and receipts | Live | Node | Site plumbing, keep |
| Marketing calendar and card editor | Live (admin) | None | Internal tool |
| Security watch and Deadman's Switch | Live | Optional AI key | Internal safety |
| Tracker and learning content | Live | GitHub API | Free, keep free |
Why kTown first
A Kisumu shop prints one QR. Any Lightning customer pays. The shop gets exact KES in M-Pesa. No wallet for the merchant. No hardware. kTown never holds money: sats go payer to Tando to M-Pesa, and Tando takes 1%. It is the only money-moving candidate that can plausibly open without a licence. The lawyer confirms.
- The direct alternative is the Tando app and API itself: zero-fee checkout, and every M-Pesa number already works as a Lightning address.
- BTCPay Server is free and non-custodial, but technical: a server and a node or provider.
- Kotani Pay offers stablecoin on and off ramp rails for builders.
- The edge is a printed QR, the exact KES promise, Kiswahili, and a human who visits the shop.
- Honest money: no revenue at launch. Later options are a flat till fee or a small add-on, lawyer first, since fee taking may change the licensing answer.
Why Ye₿ente waits
The wallet is the best tech here. It is also custodial: the server ledger holds user sats on Stanley’s node. Custodial services have their own rules under Kenya’s Virtual Asset Service Providers Act 2025, so it stays a learning wallet while that advice is taken.
- Rivals that avoid custody: Tando (non-custodial), Bitika (sats to your own wallet), Machankura (Lightning over USSD), Blink (open source, StableSats). Also in the space, depth in Kenya to confirm: Fedi, Wallet of Satoshi, Zeus, Bitnob.
- Money math: revenue is zero today (spread 0, sends free). 100 buys a month at 1,000 KES with a 1% spread is 1,000 KES a month. Growth waits for the legal answer.
- Scale math: 1,000 wallets at the 200,000 sat cap is 200M sats (2 BTC) of Stanley's outbound liquidity. The solvency check watches it. Funding it is real money.
The 30-day pilot
10 shops in Kisumu Stanley can visit on foot. One printed QR each. KES only. 15 to 50,000 KES per payment. No fees from Altradits. Kiswahili and English.
- Days 1 to 7Clearance and fixesLegal advice sought. Print flow, terms page, and abuse blocklist built and tested. Intro email to Tando sent.
- Days 8 to 14First shopsAnswers in. KTOWN on. 3 friendly shops onboarded. First real payments watched in person.
- Days 15 to 21Ten live10 shops live. Visit each. Log every failure and every support ask.
- Days 22 to 30DecideGrow (more shops, second town) or stop. The signals below decide.
Grow signals
- ✓8 or more of 10 shops still show the QR.
- ✓Repeat customer payments, not just demos.
- ✓Support under 30 minutes a day.
- ✓A shop asks for a second QR or refers a neighbour.
Stop signals
- ✕Tando blocks or rate limits the use.
- ✕Any payment lands wrong: amount or number.
- ✕Legal advice says to wait.
- ✕Support passes 1 hour a day for 10 shops.
- ✕Merchants stop showing the QR.
Stanley does himself: legal advice, the Tando email, the printing, and the shop visits. No new accounts needed. No Daraja, no Pesapal, no Bitika live key for kTown.
What stays closed
- Ye₿ente for strangers. It holds users’ sats, so it waits for legal advice.
- Buy sats and card top-up for strangers. Same reason, plus sandbox keys that must never face the public.
- Telebirr. No partner, no testing. A config flag is not a product.
- Marketing tool, ops tooling, and the contact shield as services. Single-user internals. Selling them burns the one person who keeps the node safe.
Reference data
Key dates
- 4 Nov 2025: the VASP Act commenced.
- 22 Jul 2026: the VASP Regulations gazetted (Legal Notice 134).
- 4 Nov 2026: the one year transition ended.
- 5 Nov 2026: this research reviewed.
Fees today
- Tando: 1%, kept by Tando. Bitika: 3%, kept by Bitika.
- Altradits: Buy spread 0, handle sends free. Revenue hooks exist and sit at zero.
- Tando address claim: KES 210 over M-Pesa. Contact form: $1. Tips: 10 to 5,000,000 sats.
Caps in the code
- Wallet 200,000 sats. Payment 100,000 sats. Daily send 200,000 sats. Recovery hold 1,000 sats.
- Send Mobile 15 to 50,000 KES. Bitika buy 10 to 10,000 KES per payment.
Settings that gate features (names only)
KTOWN, YEBENTE_WALLETS, YEBENTE_MPESA_BUY, YEBENTE_BITIKA_BUY, BITIKA_API_KEY, YEBENTE_BUY_PROVIDER, YEBENTE_CARD_BUY, PESAPAL_ENVIRONMENT, YEBENTE_BUY_SPREAD_PCT, YEBENTE_MAX_BALANCE_SATS, YEBENTE_MAX_PAYMENT_SATS, YEBENTE_DAILY_SEND_SATS, YEBENTE_RECOVERY_HOLD_SATS. See .env.example in the repo.
Live snapshot, 5 Nov 2026
Health checks (read-only GET)
| Check | Value |
| /api/yebente/health | Wallets on. Node connected. Send Mobile on. Buy hidden (sandbox). Card off. |
| /api/yebente/rails | M-Pesa through Tando only. |
| /api/ktown/health | Off. |
| /api/tracker | Working. Focus LND. |
| /, /yebente, /ktown | 200. |
Partner links
Sources
- Regulation: ENSafrica on the VASP Act, the 4 Nov deadline, Vellum on compliance, LawLab countdown, the CBK and CMA split, licence capital table, licence seekers, ODPC registration.
- Market: 40M Lightning addresses, Tando interview, Bitika guide, Bitika profile, Bitika outage note, Machankura on feature phones, Blink in a Kenyan village, Kotani Pay.
- Code: docs/yebente.md, docs/ktown.md, lib/ywallet.js, lib/ktown.js, lib/rails.js, lib/bitika.js, api/yebente.js, api/ktown.js, server.mjs, DEPLOY.md, in the repo.
Method
Read-only review of the main branch plus live GET checks and public sources on 5 Nov 2026. Every rival and rule links its source. Gaps say to confirm. Research, not legal advice. Questions go through Contact.