Credstone
Schedule a demo

Financial infrastructure for African
markets.

A unified platform for credit decisioning, AML, and counterparty risk. Built for the lenders and regulators running the continent's financial systems.

Schedule a demo →See the platform
credstone.console / command center livesimulated
all systems nominal·last event 8s ago

19 active tenants on the platform.

4 provisioning sagas in flight, 5 of 6 product caches healthy.

Tenants
24
across all regions
Active
19
19 of 24
Activations
41
1.7 per tenant
Users
312
across tenants

Recent activity

8 most recent
14:32ama.darkotenant.activated.atlas on Kente Trust Bank GhanaActivate
14:28service:consolesaga.completed on Marula Microfinance · prismActivate
14:21kofi.amoahuser.role.granted on Continental TrustRole
14:17ngozi.okafortenant.created on Harmattan Capital Treasury DeskCreate
14:09service:consoleservice_client.rotated on atlas → argus relayRotate
13:54thandi.mtenant.suspended on Acacia MerchantSuspend
13:42wanjiru.ktenant.activated.argus on Savanna Bank KenyaActivate
13:28service:consoleuser.invited on Lagoon Bank Nigeria · 3Create
Audit · live
14:32activatetenant t-kente14:28completedsaga 0x71a4f88e14:21roleuser u-9c3f12
The gap

Africa's $331B credit gap.

Source — IFC MSME Finance Gap · March 2025

Demand isn't the problem — legacy infrastructure is. We replace manual workflows and fragmented data with a shared intelligence layer built for regional banking.

  • Excel-based credit committees with no audit trail.
  • Sanctions screening that flags every common name.
  • Compliance teams scrambling at hour 23 on a SAR filing.
See how the platform answers it
GHGhanaBoG
NGNigeriaCBN
KEKenyaCBK
ZASouth AfricaSARB
The platform

Three products. One audit trail.

Decisioning, counterparty intelligence, fraud and AML — served from the same APIs, tuned on African data, ready for African regulators.

fraud & AML

Spot fraud in real-time.

Score every transaction in under 300ms using four parallel engines — rules, ML, sanctions, and graph.

Detect complex typologies like mule rings, structuring, and sanctions-evasion patterns. Disposition in a click: analysts see exactly which rule fired and which signal carried.

Automated SAR drafting generates jurisdictional narratives and goAML XML inside the 24-hour regulatory window. No scrambling at hour 23.

See Argus in action →
credit decisioning

Decisions that hold up.

Deploy production-grade credit scoring models trained on your proprietary lending data — no data science team required.

AutoML pipeline runs data prep, deep feature synthesis, and fairness auditing. Every score carries top-4 adverse-action reasons mapped to local requirements (e.g., Act 1052).

Replay mode: deterministically replay any decision for auditors using versioned model weights. SR 11-7–compliant model cards generate themselves.

See Atlas in action →
counterparty intelligence

Know who you're settling with.

The first neutral intelligence layer for African interbank FX and treasury markets.

0–100 risk scores for African counterparties across GHS, NGN, KES, and ZAR corridors. Settlement reliability ratings by currency-pair and time-of-day.

Rule-packs for the regulators we ship into — BoG, CBN, CBK, SARB. Prism never trades, never quotes, never the principal.

See Prism in action →
On deck

What sits underneath.

Three layers already run inside Atlas, Prism, and Argus today. Two more products are in development.

Bridge in production

Entity resolution across LEI, SWIFT BIC, and local registries. The identity layer Prism and Argus query before they score.

Forge in production

The AutoML pipeline, model versioning, and retraining workflows Atlas and Argus ship on.

Focus in production

Continuous drift, performance, and fairness monitoring. Retraining triggers when models slip.

Origin in development

Loan origination workflow — application, decisioning, disbursement — built on top of Atlas.

Vault in development

On-prem and in-region data residency. The piece you keep inside your own data center.

Oversight

Designed for oversight.

Built to clear regional central bank review.

01

On-premise by default

We don't ask you to move your data to our cloud. Deploy Credstone inside your own data center or a region-pinned VPC.

02

Hash-chained audit logs

Every decision, score, and analyst action is written to an immutable, cryptographically verifiable log.

03

Mathematical transparency

No black boxes. SR 11-7–compliant model cards and feature contribution rankings keep your risk committee in control.

04

Regional compliance

Rule-packs are pre-configured for the Bank of Ghana, Central Bank of Nigeria, and other regional authorities.

Questions

Most asked, most directly answered.

What a credit committee, a treasury desk, and a regulator each want to know before saying yes.

  1. 01Where does my data sit?
    On-premise or region-pinned. We support air-gapped deployments for national security and jurisdictional compliance. Unlike SaaS-only vendors, Credstone can live entirely inside your own data center or a private VPC in a BoG/CBN-approved region. Your data never leaves your jurisdiction unless you explicitly configure a cross-border relay.
  2. 02What do these products replace?
    We complement your existing core. Atlas, Prism, and Argus are designed for interoperability, not "rip and replace." They connect via REST APIs to your existing stack—letting you modernize decisioning, counterparty risk, or AML operations without touching your underlying ledger or core banking system.
  3. 03How do you handle 'black box' AI concerns from regulators?
    We don't use black boxes. Every Atlas score includes the top 4 adverse-action reasons mapped to local lending laws (like Ghana's Act 1052). Every Argus alert shows exactly which engine—rules, ML, sanctions, or graph—triggered the flag. We provide SR 11-7–compliant model cards as standard, giving your risk committee full mathematical transparency.
  4. 04Is there an audit trail for manual analyst overrides?
    Yes. Every action is cryptographically linked. Every score, decision, and manual override is written to an immutable, hash-chained audit log. If a risk officer overrides an Atlas decision or an analyst clears an Argus alert, the action is recorded with a timestamped hash that is mathematically impossible to alter or delete after the fact.
  5. 05How long does a full integration take?
    Provisioning is instant; integration lands in weeks. Technical tenant setup happens in seconds. A full end-to-end integration—from data ingestion and model calibration to production deployment—typically lands in 4 to 6 weeks. We build for the "architect," providing clear REST documentation and SDKs to minimize engineering friction.
  6. 06Can the system detect complex 'mule' or 'structuring' patterns?
    Specifically designed for African payment typologies. Argus uses four parallel engines, including a transaction graph, to detect mule rings, structuring, and sanctions-evasion patterns that simple rule-based systems miss. We name the typologies your compliance team is actually working on, reducing alert fatigue and false positives.
  7. 07Are the rule-packs configurable per jurisdiction?
    Your rule-packs, your way. We ship with 12 pre-configured jurisdictional rule-packs including BoG, CBN, CBK, and SARB. These are YAML-based configurations—you can update thresholds, add custom rules, or enable new jurisdictions without a code release or a vendor wait. The substrate bends; your accountability stays fixed.
  8. 08How does Prism avoid conflicts of interest in FX risk?
    Prism is a neutral intelligence layer. We never trade, never quote, and never match buyers. Most Africa-specific risk vendors are also principals, which creates an irreducible conflict of interest because they take the other side of your deal. Prism exists solely to provide the score, ensuring total neutrality in your counterparty assessment.
Talk to us

Contact us.

Talk to our team about integrating Credstone into your stack.

Replies usually within a business day.