Agricultural credit your committee can put in the file.
ACESS is an agricultural credit assessment and decision-support engine. It turns consented identity, farm, production, transaction and counterparty evidence into a graded, reason-coded assessment, a structured facility, the risk transfer required to make the exposure acceptable, and monitoring after disbursement — inside your own credit policy, through your own workflow.
~4.7%
of Nigeria's private-sector credit reaches agriculture
~US$74bn
West Africa agri-SME finance gap, per year
~US$65bn
East Africa agri-SME “missing middle” gap
Sources: BII / IFC & SME Finance Forum; ISF Advisors. The barrier is not appetite — it is evidence.
The problem
The gap is evidentiary, not ideological.
Agriculture is close to a quarter of Nigeria’s economy and a small fraction of the sectoral loan book. A credit officer cannot verify that the farm exists, cannot confirm that the borrower delivered last season, cannot distinguish a disciplined cooperative from a weak one, and cannot see whether a buyer is standing behind the transaction. Without shared, standardised, consented evidence, the rational answer is to decline.
ACESS is built to change that answer — with evidence a credit committee can document, not with a looser credit policy.
ACESS is
- ✓An agricultural credit assessment and decision-support engine
- ✓A structuring tool that sizes a facility to a production or trade cycle
- ✓A monitoring layer that watches a financed exposure after disbursement
- ✓A configuration surface for your own credit policy
- ✓An evidence trail that supports a guarantee claim or an audit — explainable by design
ACESS is not
- ×A lender. It extends no credit and holds no exposure
- ×A credit bureau. It maintains no shared register between institutions
- ×A rating agency. Its grades are not credit ratings
- ×A guarantee or an insurance policy. It indemnifies no one against credit loss
- ×A loan origination system, and no ACESS output is a credit approval
Why the independence matters
Zowasel does not use ACESS to lend on its own account. ACESS is sold to institutions as the measurement layer — which removes the conflict a bank’s model-validation function would otherwise raise. Every credit decision, its pricing, its provisioning and its regulatory treatment remain yours.How an assessment works
Gates before scores. Reason codes on every adverse outcome.
Two engines run against one subject. A Predictive Dynamic System — a machine-learning model scoring behaviour, transaction patterns and trained agricultural data — runs alongside a Dynamic Variable System, an expert-rule model built by agronomists covering farming experience, proof of prior harvest, crop suitability, location and climate-risk exposure and repayment capacity. The two combine into a unified score and an A–F grade.
Consented capture
Portal, offline Android field capture, versioned bulk upload, USSD, SMS and WhatsApp. Self-registration is never permitted.
Twelve eligibility gates
KYC/KYB, verifiable transaction, credible repayment source, fraud and duplication checks, positive transaction economics.
Two engines, one score
PDS and DVS combine into a unified score, a grade, block-by-block composition and plain-language drivers.
Facility structuring
Product fit, indicative limit, tenor aligned to the cycle, advance rate, conditions precedent, and the cover required.
Monitoring
Re-scoring, score migration, early-warning alerts, covenant tracking and stop-draw triggers after disbursement.
Only A, B or C reach a credit workflow. D, E and F are excluded but receive system-generated improvement recommendations with a target metric and an earliest re-score date. A, B or C is eligibility — never automatic approval.
What you receive
Six things a credit file does not currently contain.
A signed, defensible assessment
An Alternative Credit Assessment Report with reason codes on every adverse outcome — a document the credit officer places in the file and an auditor or guarantee provider can later rely on for a claim.
Facility structuring, not a number
Product fit from a library of fourteen designs, indicative limit, tenor, advance rate, conditions precedent, and the insurance and guarantee cover required to make the exposure acceptable.
Repayment captured at source
Where a facility is linked to the Zowasel Marketplace, verified enterprise buyers settle directly, so repayment is captured within the transaction rather than chased after harvest.
Six scorecards, including the buyer
Farmer, cooperative, merchant, enterprise and the facility itself — plus the buyer as repayment counterparty, because in transaction-linked lending the buyer's quality sets the supplier's advance rate.
Post-disbursement monitoring
Score migration since disbursement, early-warning indicators, covenant and condition tracking, trigger monitoring against insurance thresholds, and watchlist generation.
Your policy, as configuration
Weights, cut-offs, minimum DSCR, advance caps, tenor limits, concentration limits and delegated authority are configuration, not a change request. Where your policy is stricter, the stricter setting applies.
The floor beneath every configuration
No parameter cures illegality, a sanctions match, suspected fraud, fictitious trade, unverified repayment capacity, or a facility approved without authority. The eligibility gates, the prohibited-activity screens and the delegated authority record are fixed. Everything else is yours to set.Secure and covered
Risk transfer is built into the structure, not bolted on after approval.
Agricultural production risk is covariant — one drought defaults a whole zone. So cover is not an afterthought in ACESS: a facility cannot reach a disbursement-ready state without its required tiers bound, and claim triggers are defined at the outset rather than argued afterwards.
| Tier | Cover | How it applies |
|---|---|---|
| Tier 1 | Index-based crop cover | Baseline on every crop-production-linked facility. Triggered from a monitored index breach, with the claim initiated without borrower paperwork and the payout routed through the settlement waterfall to the lender first. |
| Tier 2 | Indemnity cover | Death or permanent disability of the named borrower, with the outstanding balance settled to the lender. |
| Tier 3 | Warehouse cover | Required for inventory finance and warehouse receipt finance, against physical loss and theft of stored stock. |
| Tier 4 | Pluggable cover | Credit insurance on receivables, goods in transit, equipment and asset cover, and multi-peril crop cover, applied by product. |
Guarantees, stacked and tracked
Guarantees are handled as a registry with stacking. A facility can carry an institutional guarantee and a cooperative first-loss position at the same time, with the residual uncovered exposure displayed and the cost of cover embedded into the facility economics.
Two structures, deliberately different
Trade finance substitutes the offtaker’s credit for the borrower’s, and self-liquidates against an identified asset or payer. Production finance cannot make that move, which is why the guarantee and insurance layer sits mostly on the production side.
To be precise about what this is.
ACESS is not a guarantee and not an insurance policy, and it indemnifies no one against credit loss. What it does is identify, recommend, bind and monitor the cover that a third-party guarantor or insurer provides, and assemble the evidence that supports a claim.Three ways to run it
Portal, API, or entirely inside your own environment.
Dashboard
Ten role-specific views, each answering a decision its user is accountable for. Most phase-one pilots start here, with no integration work.
- ·Underwriting view — a two-minute decision screen
- ·Risk and portfolio view, with vintage curves and an override register
- ·Programme funder view for DFIs, with reach, additionality and fairness evidence
- ·Guarantee and insurance provider views with live trigger monitoring
- ·Model governance view against published gates, pass or fail explicit
- ·Permissioned, read-only regulator and auditor view
API
The score and the report appear inside your existing loan-origination or core-banking screen. Read the specification as a contract: it is delivered and tested with you in sandbox before any production dependency is created.
- ·OAuth 2.0 or issued keys over TLS 1.2+, role- and tenant-scoped
- ·Signed webhooks with replay protection; idempotency keys on every state change
- ·Immutable score runs with an input snapshot hash and a reproducible explanation
- ·Versioned paths and a minimum twelve-month deprecation notice
- ·Batch and secure file transfer where change control will not permit an interface
On-premise
Single-tenant on your own hardware, where personal data never leaves your network unless you configure it to. Built for regulated and government-linked institutions.
- ·Versioned installable artefact with an offline installation path
- ·Defined degraded-mode operation for every unreachable dependency
- ·Model updates without data egress
- ·Quarterly update packs plus critical patches, on your approval
- ·In-network interface, message queue or batch to core banking
99.5% availability
Measured monthly, with service credits.
Sub-second at median
For a single assessment on a data-complete subject.
P1 within 4 hours
P2 within 8 business hours, P3 within 2 business days.
RPO 15 min · RTO 4 hrs
Default recovery objectives, negotiable per institution.
Who ACESS is for
Five institution types, one measurement layer.
Commercial banks
Originate and monitor agribusiness trade tickets above microfinance limits — purchase-order finance, receivables, inventory and structured commodity finance.
A finance-ready pipeline with documentation assembled, portfolio visibility across a value chain, and reporting that supports development-finance and buyer requirements.
Microfinance banks
Assess and structure in-scope agricultural facilities within micro-loan and single-obligor limits, and run programme accounts for a served network.
Deposit and account growth, structured evidence per facility, fee and float income, and access to guarantee tiers that reach 75 per cent for smallholders and cooperatives.
Telcos & PSBs
Distribution, wallets, servicing and consented behavioural signals alongside a licensed credit partner.
Wallet activation and transaction volume in rural markets, and participation in agricultural credit economics without lending.
Insurers
Attach index, indemnity, warehouse and transit cover to a scored, monitored exposure.
A pooled, screened book with live trigger monitoring, and claims initiated from monitoring events rather than from paperwork.
DFIs & guarantee providers
Make guarantee and risk-sharing capacity usable by giving partner lenders the means to originate, assess and monitor.
A deployment channel with transaction-level reporting, evidence packs that support a claim, and impact and fairness measurement from the same record.
For DFIs and guarantee providers
ACESS turns risk-sharing and guarantee capacity into deployable credit. It supplies the transaction-level assessment, monitoring and claim-ready evidence that let a DFI stand behind agricultural lending it could not otherwise reach — and the loss data to price it.Governance, validation and our honest position
Supervisory guidance gives vendor models no exemption. Neither do we.
Your validators must be able to test ACESS to the standard applied to your own internal models. The documentation is written for that rather than around it — and the deployment thresholds are published before the result is known, rather than after.
Designed and tested under a JICA-supported programme
ACESS was designed and tested from 2022 under a JICA-supported programme on credit scoring for smallholder farmers in Nigeria, with real lending delivered through financial-institution partners.
Independent third-party audit completed
Data Aid — engaged under TechnoServe’s FCDO-funded CASA Plus programme — completed an independent design, logic, control and validation-readiness audit aligning ACESS with Basel credit-risk parameters and IFRS 9 expected credit loss concepts.
What your model risk function receives
Discriminatory power reported separately on development, out-of-sample and out-of-time populations; calibration by score band; back-testing and a challenger model; a data dictionary with consent basis per field; an exportable reason-code catalogue.
Published deployment gates
AUC ≥0.70 (0.75 for bank deployment). Gini ≥0.40 (0.50 for bank). KS ≥0.30. Population stability <0.10 stable. Hosmer–Lemeshow >0.05. These are the gates. They are not results.
Security and data protection
DPA executed before processing; TLS 1.2+ in transit, AES-256 at rest, row-level tenant isolation, SSO/MFA. Zowasel does not hold ISO 27001 or SOC 2 — SOC 2 Type II is the first certification target.
Ready for / not ready for
Ready for: parallel assessment alongside your existing process on live applications; structuring and standardising underwriting evidence; sizing and shaping facilities for committee review; building the outcome record calibration requires.
Not ready for: autonomous production credit approval; risk-based pricing derived from model probabilities; automatic limit setting without lender review; or making origination dependent on the interface before joint validation.
How an engagement starts
We do not ask you to trust ACESS. We ask you to test it.
Three phases, with movement between them by written gate criteria agreed in advance — not by a milestone anyone can declare.
8 to 12 weeks · Parallel assessment
ACESS runs alongside your existing process, on live applications
Both assessments are recorded and compared. Models are calibrated to your portfolio and signed off by your risk function. Sandbox, dashboards, data templates and the security review all run in this window.
Your risk: none. No decision is made on an ACESS output.
12 to 24 weeks · Limited production
ACESS informs decisions within an agreed scope
A defined product, segment, ticket band or branch set, with the full workflow, reporting and integration in use.
Your risk: contained and reversible within a defined scope.
Balance of term · Full production
Use across the agreed portfolio
Integrated into the origination or core-banking environment, with monitoring and portfolio reporting live.
Your risk: normal vendor risk.
Commercial shape
Pricing is set per institution in a term sheet, and we will walk your finance team through it on the call. Priced per decision, not per seat — a scoring tool priced per user is a tool an institution rations. There is a published concessionary tier for institutions substantially funded by development or donor capital.Request a meeting
Tell us about your institution. We will bring the right people to the call.
Eight minutes. A Zowasel officer responds within two working days with a proposed agenda and the materials relevant to your institution type. No obligation, and nothing here is an application for anything.
What happens next
Three steps, and we do most of them.
We call you
Within two working days, to confirm scope, attendees and the agenda — and to tell you which materials need an NDA first.
We meet
A demo, or the ninety-minute institutional design workshop with your credit, risk and digital teams, scoped against your own portfolio, commodities and geography.
You decide on a phase one
If it fits, we scope an eight-to-twelve-week parallel assessment where no decision is made on an ACESS output. If it does not, we will say so.