Skip to content
PesaRails
Karibu · The credit rails for African MFIs

Approve the right loans.
Collect on time.

PesaRails is the credit infrastructure behind Kenya's microfinance institutions: AI-driven scoring, a WhatsApp loan assistant, and a compliant core MIS, unified on one platform.

CBK-aligned
Kenya DPA compliant
M-PESA native

PesaRails

online

👋 Karibu Wanjiru. Would you like to apply for a loan today?

09:42

Yes, KES 25,000 please

09:42 ✓✓

Approved! KES 25,000 on its way to your M-PESA.

Score: 742 · 30 days · 6% p.m.

09:43

Asante sana! 🙏

09:43 ✓✓

Message
The category

Shopify, but for microfinance.

Running an MFI in 2026 means stitching together a core banking system, a credit scoring service, a WhatsApp BSP, an M-PESA integration, a fraud tool, a bureau connector, and spreadsheets. Six vendors. Six contracts. Six places your data can leak.

PesaRails collapses that stack into one. Launch a new lending product in an afternoon. Run compliant loan books across branches, regions, and countries on a single tenant, with the controls CBK expects already baked in.

We don't lend money. We sell the software to the organizations that do.

Multi-tenant from day oneSchema-isolated per MFIBilingual EN / SWAPI-firstKenya-specific rails
The gap we close

Three-day loans in a three-minute world.

Kenya's microfinance operators are running 2026 businesses on 2006 infrastructure. Here's what that costs them, and the borrowers they serve.

3 days

Average branch-to-M-PESA lead time on a manual loan

8–12%

Default rates at MFIs without systematic scoring

KES 810M

M-PESA fraud losses in 2024, a 344% YoY jump

6–12 mo

Typical Temenos or T24 implementation time

Why MFIs choose PesaRails

Run the business, not the software.

Every tool a loan officer needs, from KYC to collections, on one platform designed for the Kenyan credit ecosystem.

Faster

Approve in under 90 seconds

AI scoring pulls M-PESA history, IPRS identity, and CRB bureau data to return a verdict before the borrower closes WhatsApp.

Safer

Lower default rates

Behavioural features like transaction patterns, gambling signals and income stability catch risks that traditional rule-based scoring misses.

Compliant

Audit-ready by default

CBK adverse-action codes, DPA consent capture, and immutable audit logs built into every workflow, not bolted on.

How it works

Three products. One platform.

Adopt one, or all three, and they talk to each other out of the box. No middleware, no glue code, no six-month integrations.

01

The frontline

WhatsApp loan chatbot

Customers apply, upload statements, check balance, and repay, all in WhatsApp. No app downloads, no training, no support tickets.

  • Bilingual EN/SW flows
  • M-PESA statement parsing
  • OTP-verified consent
02

The brain

Credit scoring engine

A tenant-configurable scoring model blends bureau data, mobile money behaviour and loan history into a 300–850 score with adverse-action reasons.

  • CRB + IPRS + M-PESA signals
  • Explainable adverse-action codes
  • A/B test new models safely
03

The back-office

Core MIS

Built on Apache Fineract and extended for Kenyan MFIs: disbursements, collections, reporting and audit trails, all in one dashboard.

  • Apache Fineract foundation
  • CBK-ready reports
  • Role-based permissions
Deployment models

Meet your stack where it is.

Three ways to onboard, from greenfield to full replacement of a legacy core. Every model ships with the same scoring engine, the same WhatsApp assistant, the same audit trail.

Most popular

No existing core banking

Standalone

PesaRails is your entire platform. We provision everything: loan book, scoring, WhatsApp, M-PESA, reports.

Ideal for

New MFIs · SACCOs going digital · Lending pilots

Managed Apache Fineract

Core MIS

We provision a dedicated Fineract instance for your MFI and wrap it with scoring, WhatsApp, and M-PESA rails.

Ideal for

Established MFIs modernizing · Community banks

T24, Temenos, Musoni, in-house

External CBS

Keep your existing core. PesaRails plugs in via webhooks and becomes your scoring + WhatsApp + fraud layer.

Ideal for

Tier 2 banks · Digital lenders · Scale-ups

The platform, in detail

Built for loan officers, not for developers.

Credit scoring engine

A score you can explain to a regulator, and to a borrower.

Every verdict comes with up to four adverse-action codes mapped to CBK guidance. No black boxes, no surprises in audits.

  • 300–850 FICO-style band, calibrated to KE cohorts
  • M-PESA, bank statement, CRB, and IPRS signals
  • Per-tenant model weights, no one-size-fits-all

Applicant

Wanjiru Njeri

KES 25,000 · 30 days

Score

742

Very Good

300500700850

Signals driving this score

  • M-PESA inflow consistency (24 mo)+182
  • Zero CRB adverse listings+95
  • 2 prior loans repaid on time+60
  • High weekend gambling spend-45
Under the hood

An ensemble a regulator will actually sign off on.

XGBoost, LightGBM, Random Forest, and Logistic Regression vote on every applicant. SHAP explanations attach to every decision, so credit committees, auditors, and the borrower get the same reason codes.

Owner characteristics

40%

M-PESA behaviour, age, business tenure

Business characteristics

30%

Sector, years in operation, revenue stability

Loan characteristics

15%

Amount-to-income ratio, tenor, purpose

Behavioural indicators

15%

Savings pattern, balance stability, gambling exposure

Score bands & policy

FICO-style 300–850 scale, calibrated on Kenyan cohorts

Credit score bands, risk levels, and lending policy actions
ScoreRiskPolicy action
750–850ExcellentAuto-approve · up to 10× monthly income
650–749LowApprove · up to 5× monthly income
550–649ModerateApprove · up to 2× monthly income
450–549HighManual review queue
300–449Very HighAuto-decline with adverse-action notice

Adverse-action codes, ready to mail

Every decline ships with up to four CBK-aligned reason codes (AA001 – AA024). The notice generator drops a PDF into the customer's WhatsApp the moment the decision is final.

End to end

From hi to M-PESA in under 30 minutes.

A 24-state conversation engine orchestrates the entire loan lifecycle inside WhatsApp. No app. No branch. No paperwork.

  1. 0:00

    Step 1

    Hi

    Customer messages the MFI on WhatsApp. Language prompt and terms land in under 10 seconds.

  2. 0:02

    Step 2

    KYC

    Name, ID front, ID back, selfie, next-of-kin, guarantor, GPS, business info, all via WhatsApp buttons.

  3. 0:12

    Step 3

    Statement

    M-PESA PDF uploaded. Ensemble model + LLM return a 300–850 score with reason codes.

  4. 0:15

    Step 4

    Apply

    Amount, tenor, purpose. Amortization schedule surfaces before the customer accepts the digital agreement.

  5. 0:17

    Step 5

    Approval

    Auto-approve under threshold, or route through branch → region → HQ with SLA timers.

  6. 0:28

    Step 6

    Disburse

    M-PESA B2C fires. Receipt returns to WhatsApp. The loan goes live on the MIS.

Fraud · AML

Eight layers between your loan book and bad actors.

KES 810M was lost to M-PESA fraud in 2024, a 344% jump year on year. PesaRails stacks eight defensive checks across every application so the losses don't land on your balance sheet.

01

Circular transaction rings

A → B → C → A within 24–48 hours gets flagged before scoring even runs.

02

Statement tampering

PDF forensics catch font swaps, edited balances, stitched-together statements.

03

Velocity anomalies

Transaction counts >3× historical average pause for manual review.

04

Income inflation

Large unexplained deposits right before an application drop the score automatically.

05

Duplicate statements

Image and document hashing prevents the same statement being reused across loans.

06

Date manipulation

Timestamp re-ordering and out-of-sequence dates raise a hard block.

07

Cross-person collision

Applicant ID ≠ next-of-kin ID ≠ guarantor ID, enforced at KYC time.

08

Image hash dedup

Re-uploaded selfies and IDs across applications are caught by perceptual hash.

Integrations

Plugged into the rails you already use.

Kenya-specific connectors out of the box. No consultants. No 6-month onboarding.

M-PESA

M-PESA

Daraja STK push & C2B

WhatsApp

Cloud API · verified

Apache Fineract

Core banking engine

metropol

CRB Metropol

Bureau score pulls

IPRS

National ID verify

Google Cloud

Sovereign KE region

Who runs on PesaRails

Built for the full shape of African lending.

From a twenty-person SACCO to a Tier 2 bank: same platform, same rails, different knobs.

  • SACCOs

    ~15,000 registered in Kenya

    Member self-service, group lending, automated dues

  • Deposit-taking MFIs

    14 CBK-licensed institutions

    Replace legacy cores. Cut loan decision time 99%.

  • Credit-only MFIs

    2,000+ small lenders

    Go digital end-to-end without hiring an engineering team.

  • Digital lenders

    100+ Kenya-native fintechs

    Scoring API + fraud stack, with your UX on top.

  • Community banks

    30+ Tier 2 & Tier 3

    Add a WhatsApp front door without ripping out your core.

Versus

The fastest path to a live loan book.

Feature comparison: PesaRails vs legacy MIS, in-house build, and spreadsheets
Capability

PesaRails

You, with us

Legacy MIS

T24 · Temenos · Musoni

Build in-house

2 engineers · 18 months

Spreadsheets

How most MFIs still run

Time to first loanAfternoon6–12 months12–18 monthsSame day
WhatsApp originationNativeAdd-onBuild it yourselfNone
AI credit scoringIncludedNoneBuild it yourselfNone
M-PESA B2C & C2BNativeCustom integrationBuild it yourselfManual
CBK adverse actionAuto-generatedManualBuild it yourselfManual
Fraud detection8 layersNoneBuild it yourselfNone
Typical annual costKES 720K to 3MKES 10M+KES 30M+ (salaries)Staff time
Security · compliance

Built for regulators. Safe for borrowers.

PesaRails is architected from the ground up around Kenya's Data Protection Act, CBK digital credit guidelines, and global banking security standards. Every tenant gets isolated data, role-based access, and end-to-end audit logs.

CBK

Digital Credit Providers

DPA

Kenya Data Protection Act

ISO 27001

Information security

SOC 2

Type II, in progress

  • Tenant-isolated data

    Each MFI gets its own schema. API keys cannot cross organizational boundaries. Ever.

  • Immutable audit trail

    Every approval, override, and repayment is logged with actor, timestamp, and IP, ready for CBK inspection.

  • SSO · RBAC · MFA

    Role-based permissions mirror CBK segregation-of-duty rules. SAML SSO for enterprise tenants.

  • DPA-compliant consent

    Consent captured at loan origination, versioned, and surfaced on every data-use request.

  • Data sovereignty

    Hosted in Google Cloud’s Johannesburg & Nairobi regions. Your customer data never leaves the continent.

Pricing

Priced for Kenyan P&Ls.

Transparent tiers. No per-seat surprises. Switch plans when your loan book does.

For pilots & small SACCOs

Starter

KES 60K/month
Talk to sales
  • Credit scoring API · 100 statements/mo
  • 1 admin seat
  • M-PESA sandbox
  • Community support
  • Email-only onboarding
Most MFIs start here

For most Kenyan MFIs

Growth

KES 250K/month
Book a pilot
  • Scoring + WhatsApp chatbot
  • 10,000 statements/mo
  • 5 admin seats
  • M-PESA production rails
  • CBK reporting pack
  • 8×5 support

Tier 2 banks & digital lenders

Enterprise

From KES 1.5M/month
Talk to sales
  • Full platform incl. Core MIS
  • Unlimited statements
  • 50+ admin seats
  • Dedicated Fineract instance
  • Custom integrations (CRB, IPRS)
  • 24×7 named CSM

Prices are indicative · final pricing after scoping call

Common questions

Questions every MFI asks.

Didn't find yours? Email info@pesarails.com and we reply inside a business day.

Is PesaRails a lender?

No. We are a B2B platform: the software that microfinance institutions use to run their lending operations. We never touch borrower capital.

How long does onboarding take?

A Starter tenant can be live in a single afternoon. Growth tenants typically pilot in 2–3 weeks, including Fineract provisioning and M-PESA go-live. Enterprise with custom integrations runs 4–8 weeks.

Is my data isolated from other tenants?

Yes. Every tenant gets its own Postgres schema with row-level security, its own AES-encrypted document store, and its own set of API keys. No request can ever cross an organizational boundary.

Which channels does the chatbot support?

WhatsApp Cloud API is primary. USSD is on the near-term roadmap. Free-text natural language understanding (including Sheng) is planned. Today the flow is button-driven, which is more reliable for regulated lending.

Can I keep my existing core banking system?

Yes. The External CBS deployment model lets you keep Temenos, Musoni, or an in-house system. PesaRails plugs in via webhooks and becomes your scoring + WhatsApp + fraud layer on top.

What about CBK licensing?

PesaRails is technology, not a lender, so we don’t require a digital credit provider license ourselves. Your MFI’s existing CBK license covers the lending activity, and our compliance surface (adverse-action notices, audit logs, DPA consent capture) plugs straight into your regulatory reporting.

Tuko tayari

Put your loan book on rails.

Book a 30-minute product walkthrough. We'll set up a sandbox tenant before the call ends.

Book a demo

See it running on your own numbers.

Thirty minutes, no slides. We walk through scoring, WhatsApp origination and the admin console, then set up a sandbox tenant you can keep poking at afterwards.

Response time
Within one business day
Office
Nairobi, Kenya

No sales sequence, no newsletter. One reply from a human.