Let a driver finance their motorcycle through Drivra. Who else is already doing this and how, what Nepal's regulator actually requires, the full system architecture, the driver-facing journey, and how credit gets decided with no external bureau in year one.
RYD Nepal (rydnepal.com) rents Hero Super Splendor 125cc motorcycles to Kathmandu gig drivers today. Founded 2025, self-funded, 120+ bikes, claims 500+ active riders, single-SKU fleet. Only one of their three plans actually leads to ownership (Pro Monthly, Rs 7,000/week, after 1.5 years), which is worth noting since most of their marketing describes "own the bike in 18 months" as if it applies broadly. KYC is driving license + citizenship copy only, under 30 minutes, zero down payment, no app (web form + WhatsApp + in-person).
| Plan | Price | Leads to ownership? |
|---|---|---|
| Weekly | Rs 5,600/week | No |
| Pro Monthly ("Best Value") | Rs 7,000/week | Yes, after 1.5 years |
| Prepayment | Rs 21,000/month (Rs 700/day, cheapest per-day rate) | No |
Free service every 1,500km, 24/7 breakdown line, a replacement bike within 30 to 40 minutes if repair runs long, and a welcome "Sagoon kit" (helmet, phone mount, raincoat) plus a Rs 1,500 fuel coupon on pickup. Missed-payment mechanics are not disclosed anywhere public, marketed only as "no penalty fees, no credit-bureau drama." Trust signal: only three self-hosted testimonials, zero independent reviews found anywhere (no Trustpilot, Google reviews, Reddit), so their credibility is entirely self-curated, itself an opening.
| Company | Market | Model | Notable mechanic |
|---|---|---|---|
| M-KOPA | Kenya, Uganda, Nigeria, Ghana, S. Africa | PAYG motorbike financing, KES 10 to 25k deposit + daily installments | Missed payment extends the term with no penalty; can return the bike and walk away clean with deposit refunded. Strongest trust-building policy found in this research. |
| Tugende | Uganda, Kenya | Lease-to-own boda financing, about $120 down, about 82 weekly payments | Requires a family guarantor plus 3-week training before financing, a soft underwriting layer. Builds its own proprietary repayment dataset via its own payments platform ("PayWay") rather than external bureau data. |
| Watu Credit | East Africa, plus Mexico/Brazil | Lease-to-own, under 2hr approval, "no credit history needed" | Requires 6+ months of active mobile-money history as the alt-credit proxy. GPS remote-immobilize on default, 7-day redemption window. Real reputational problems: Facebook impersonation scams, consumer-advocacy complaints about collection conduct. |
| Moove | Africa, UK, India (Uber's financing partner) | "Ride-to-own," up to 95% financed, 24 to 48mo tenure | Auto-deducts weekly repayment directly from Uber trip earnings. Nigeria driver protests over insufficient take-home pay and aggressive in-person debt collection, a clear cautionary case. |
| Grab (GrabScheme) | Indonesia (motorcycles) | Platform = distribution/data, bank-backed multifinance company = balance sheet | Grab partnered with PT Central Santosa Finance (a Bank Central Asia subsidiary) rather than self-financing, the clearest precedent for a tech-layer-plus-licensed-partner structure. Separately, Nikkei Asia reported Grab charging over 230% effective APR on short-term in-app loans to Philippine drivers. |
| Uber Vehicle Solutions | USA | Curated marketplace of third-party rental/lease partners | "Aggregator, not lender" pattern; Uber never originates financing itself. |
| Ola Fleet | India | Lease-to-own, evolved over several program iterations | Active legal disputes with drivers over alleged vehicle confiscation and unfulfilled ownership promises, a direct warning on keeping ownership-transfer terms unambiguous. |
| Bangladesh banks (via Pathao) | Bangladesh | Banks underwrite motorcycle loans using ride-share earnings as qualifying income | Pathao itself runs no captive financing arm; banks accept 6 months of platform-earnings statements as income proof. The lighter-touch, no-captive-lender alternative. |
| Bajaj Finance / Hero FinCorp | India | General retail two-wheeler finance, up to 95 to 100% financed | No purpose-built gig-driver underwriting product; general bureau-based retail finance still dominates in India. A possible market gap rather than a proven approach. |
| Product | Shape | Terms |
|---|---|---|
| eSewa easyloan | In-wallet lending, wallet-history-based eligibility, 6+ months active usage required | Rs 15,000 to 200,000, 17 to 18% annual interest, 2% penal interest |
| Khalti + NIC Asia "Super Chamatkarik Loan" | Khalti is a pure front-end/PSP; an NIC Asia rep manually processes within 48 hours off-app | Hybrid, not fully self-serve |
| Foneloan (F1Soft/eSewa's parent group) | White-labeled digital lending embedded across 11+ Nepali banks' apps | Payday, EMI, and a BNPL product ("HamroBNPL"), the closest Nepali analogue to embedding financing inside an app people already use daily |
The common thread across all three: wallet/account transaction history substitutes for a credit bureau score, and underwriting is handed off to a licensed bank/PSP partner behind the scenes, precisely the shape this product should take.
Nepal Rastra Bank (NRB) regulation, not engineering choice, determines what's actually buildable. Under BAFIA, only a licensed BFI may originate or hold consumer loans, a pure tech company cannot lend directly (BAFIA 2073, Chapter 6, Section 31). Vehicle financing sits under a Hire Purchase Company license: NPR 300M minimum paid-up capital, a 60% LTV cap (at least 40% down payment), per the Sixth Amendment (roughly August 2026, only weeks before this research).
Recommended for design purposes: Path A. Closest fit to the standalone-service decisions already made, cleanest regulatory footing without a NPR 300M capital commitment. This is a legal decision for Nepali financial-regulatory counsel, not resolved here.
Every step is new, part of the Financing build. Gray throughout: nothing here is built yet.
Standalone service: own repo, own Postgres database, reusing Upahar's proven stack (FastAPI + SQLAlchemy 2.0 + Alembic + Jinja2/HTMX admin), but its own deployment and security perimeter, since it holds KYC documents and financial data. It reads driver identity and performance data from the canonical identity layer described in Architecture, never direct database access into Upahar or Express.
Driver auth: phone + SparrowSMS OTP, issuing a JWT scoped to a distinct
financing_driver audience, deliberately not SSO-trusting Upahar's session for v1
to keep security boundaries independent. Admin auth mirrors the bcrypt+JWT pattern with more
granular roles: KYC reviewer, underwriter, collections/ops, owner/compliance.
At this scale (a few thousand drivers x about 5 documents), storage/egress cost is trivial either way. The deciding factor is integration effort: Supabase Storage wires directly into Postgres RLS. This must be its own Supabase project, never the one Upahar and GPS Dashboard already share.
Per-driver verification cost is roughly $0.07 (Textract AnalyzeID for 2 documents + Rekognition liveness + face-compare) versus $2 to $7 per verification for global vendors like Jumio/Onfido. DIY is roughly an order of magnitude cheaper; the trade-off is owning fraud-detection tuning in-house, acceptable when a human still signs off on every KYC approval regardless.
Driver-facing web portal (not a second app), linked from Upahar. Admin console adding new tabs to the existing page+endpoint convention. A field-agent tool to record cash payments collected in person, matching how Nepali MFIs actually collect.
| Table | Purpose |
|---|---|
drivers | Identity, linked to the canonical identity layer via phone |
kyc_documents | Uploaded documents: citizenship/NID/license/PAN/selfie, review status |
guarantors | Cold-start co-signers |
vehicles | Asset registry, deliberately separate from Upahar's driver-vehicle columns, which reflect only the driver's current Yango-assigned vehicle |
loan_products | Configurable plan terms: tenure, down payment %, interest, penalty rate |
applications | Origination workflow state |
loan_accounts | The active agreement, including lender_of_record which absorbs the legal-path decision without blocking the build |
repayment_schedule | Amortization |
ledger_transactions | Append-only ledger, mirrors Upahar's PointTransaction shape |
delinquency_events | Collections state: current/watch/substandard/loss |
audit_log | Every state change and admin action |
Every new table's migration revokes Supabase's default
anon/authenticated PostgREST grants, per Upahar's existing hard rule, now
a standing checklist item for any new Supabase-backed service across the whole platform.
Every thin-file lender researched (Watu, Tugende) starts here: a rules-based scorecard, not an ML model, since there's no proprietary repayment history yet to train on.
Inputs available day one, all from Upahar: tenure, lifetime trips/gross and their consistency, current/longest streak, active status, tier/XP. Once the identity layer exists, also from Express: delivery tenure and reliability, for drivers who do both.
Cold-start handling: a guarantor requirement and/or higher down payment for drivers with little history, mirroring Watu's guarantor model and Tugende's reference-plus-training approach, rather than an outright decline. Decision flow: hard gates (KYC complete, active status, minimum tenure) then a weighted score, then approve / decline / refer-to-manual-review, every decision and override audit-logged.
Confirmed directly against each provider's own developer documentation: none of eSewa, Khalti, or Fonepay support true unattended auto-debit. eSewa's closest product still requires the customer to actively enter their token at eSewa on every transaction. Khalti's docs show no subscription/mandate capability at all. Fonepay's merchant API docs on this specifically weren't locatable, treat as unresolved rather than confirmed absent. ConnectIPS/NCHL supports mandate-based direct debit, already used for hire-purchase EMI collection, the best path to real automation for drivers with a bank account.
ConnectIPS mandate automation is an opt-in later step, once the manual flow is proven. Cash/agent collection is the fallback via the field-agent tool.
Fee comparison: ConnectIPS is flat and cheap (Rs 2 to 8/transaction regardless of amount, NRB bars passing this to customers); Fonepay QR is free for merchants; eSewa/Khalti card-based loading runs 1.75 to 5%. Default drivers toward ConnectIPS or Fonepay QR; support eSewa/Khalti as fallback, not the primary path. This v1 flow matches how the market leader (Hulas Finserv) already operates.