15. Final recommendation

This is the section that exists to answer the founder’s question: what should I actually do?

Sub-pages

The blunt answer

1. Business model

Hold an NBFC licence. Run own-book + one bank co-lending partner under CLM-1. Layer LSP fees from one other partner. Distribute via CA networks, Tally / GST data, and one anchor programme.

Avoid: marketplace, BNPL, app-only consumer lending, secured products at MVP, multiple partners at once, managed-credit / AIF structures, productisation before book is proven.

2. Regulatory path

File NBFC-ICC CoR application in Phase 0.
If licence is in process, run as LSP for a partner NBFC in the interim — keeps building the platform without sitting idle.
Compliance officer + Internal Audit established before Phase 1 production.
Quarterly Risk + Audit Committee from day 1.
Annual external IT audit + pen test mandatory.

3. Initial customer segment

SME / MSME working-capital borrowers:

2+ years vintage.
GST turnover ₹40 lakh – ₹10 Cr.
Industries — wholesale / distribution / services / manufacturing (non-discretionary).
Geographies — 5 – 8 tier-1 / tier-2 cities at MVP.
Acquisition wedge — CA networks, Tally / Zoho Books partners, one anchor programme.

4. First lending product

Revolving working-capital line.

Ticket ₹20 – 50 lakh.
Tenure 60 – 180 days per draw.
APR 20 – 22% blended.
Bullet repayment per draw; line revolves on repayment.
Annual / quarterly limit review.

5. Vendor stack at MVP

Layer	Vendor
LOS + LMS core	Own build (lean) or CloudBankin / Finezza / Biz2X
KYC + KYB	Karza or Signzy or IDfy
V-CIP	IDfy or Signzy
AA	Setu or FinBox
BSA	Perfios
GST	Cygnet or Vayana
eSign + eStamp	Leegality + Digio
NACH + UPI AutoPay	Digio + Razorpay
Payments + payouts	Razorpay / Cashfree
SMS + WhatsApp	Gupshup
Dialer	Exotel / Knowlarity
Bureau	All four (via aggregator)

6. What to build internally

Credit policy + rule library.
Co-lending allocation + settlement.
Repeat-borrower scorecard.
CA / Tally distribution UX.
Portfolio analytics + warehouse layer.
LMS (own-build by year 2 if started on vendor).

7. What to buy

KYC / V-CIP / CKYC / AA / BSA / GST / eSign / eStamp / NACH / UPI / payments / SMS / WhatsApp / dialer.

8. What to avoid

Multi-channel sales without underwriting capacity.
Multi-product launch at MVP.
Multi-partner at MVP.
Multi-tenancy before SaaS productisation is on the roadmap.
App-only borrower acquisition (loses to the CA-led wedge).
Aggressive recovery practices (regulatory and reputational landmines).
Secured products at MVP (operationally complex, slow capital turn).

9. Sequencing

Months 1 – 3: Phase 0 (licence + setup). Sub-page: 90-day plan.
Months 4 – 9: Phase 1 (lending MVP, first 100 disbursals).
Months 9 – 15: Phase 2 (co-lending MVP).
Months 15 – 24: Phases 3 + 4 (SaaS portal + advanced underwriting).
Months 24+: Phases 5 – 7 (portfolio intelligence + multi-partner + scale).

Sub-page: 12-month roadmap.

10. Metrics that matter

Weekly: disbursements, applications, conversion at each stage, decision TAT, first-loan default rate, bucket roll-rate, vendor SLA. Sub-page: Weekly metrics.

11. What kills this business

See Fatal mistakes.

The reality of execution

This blueprint is 35 – 50 person-months of engineering + ops + risk + compliance, ~₹15 – 25 Cr of platform investment in year 1, and a real shot at ₹100 – 200 Cr book in 18 – 24 months if execution is tight.

The biggest delta between platforms that make it and don’t is discipline — saying no to scope creep, holding to the wedge, building credit muscle ahead of sales pressure, treating compliance as first-class.

Nothing in this spec is a guarantee. Everything in it is opinionated. Apply judgment.