Skip to main content
Product

An identity verification platform built for Africa’s regulated markets

Naya ID brings document reading, face biometrics, AML screening and human review into one flow. Each block works alone or combined, integrates through a single API and produces, for every decision, an evidence file you can hand to your regulator.

43 s
median time from capture to decision
12 countries
whose official documents are recognised natively
99.1%
of cases cleared without manual intervention
4 types
of documents: ID card, passport, licence, and the CIP in Benin
How a verification runs

Four steps, one minute, a documented decision

The flow is the same whether your customer verifies from your app, a secure link or a branch counter. Every step is timestamped and kept with the case file.

  1. Guided captureThe customer photographs their document and their face from their phone. Framing, sharpness and glare are checked live: unusable shots are rejected before upload, which avoids retries and drop-offs.
  2. Document reading and authenticationThe document is identified, its fields extracted by our OCR engine trained on the zone’s documents, and its integrity checked: consistency of fonts and backgrounds, visible security features, MRZ reading and check-digit validation where the format provides them.
  3. Checking the holderThe captured face is compared with the document portrait, and passive liveness rules out screen photos, replayed video and masks. Sanctions, PEP and adverse-media screening runs in the same call.
  4. Decision and evidence fileYour rules decide: approved, refused or routed to human review. The outcome is delivered by webhook and the file (documents, scores, log of every check) stays available and exportable.
The four blocks

A modular foundation, switched on according to your risk exposure

You set the level of control that fits your product, your regulator and each customer’s risk profile. The blocks share one API, one decision repository and one evidence file.

Document verification

Recognise and authenticate the documents actually in circulation.

Models are trained on the documents of the twelve covered countries, across their generations and states of wear, including laminated, reprinted or older documents that generic engines reject. Extracted fields feed straight into your customer record.

ID cards, passports and driving licences from the 12 covered countries, CIP in Benin
Forgery detection: portrait substitution, altered fields, reprints
MRZ reading and check-digit validation where the format provides them
Structured OCR extraction: names, date and place of birth, number, validity dates
Complementary documents on demand: vehicle registration, insurance certificate, business documents, read by OCR

Face biometrics

Establish that the document holder is actually present.

The match between the captured face and the document portrait is completed by passive liveness: no gesture is imposed on the customer, which lowers drop-off and works on entry-level handsets and unstable connections alike.

Face-to-document match with a score and configurable thresholds
Passive liveness: photo, replayed video, mask and screen attacks blocked
1:1 re-authentication of a known customer, in two seconds
Works without a continuous connection, with capture resumed where it stopped

AML screening

Document the risk before the relationship begins.

Screening queries international and regional sanctions lists, politically exposed persons and adverse media. Matches are ranked and explained, so your compliance team decides on evidence rather than on an isolated score.

International sanctions and the regional lists relevant to the zone
Politically exposed persons and their close associates
Adverse media, with source and publication date
Periodic re-screening of your customer base, alerts on status changes

Review console

Give your teams what they need to decide, and to justify it.

Edge cases arrive in a review queue with everything collected. Every manual decision is traced, attributed to a named agent and motivated. Conversion, rejection and fraud statistics feed the steering of your flows.

Prioritised review queue with explicit hold reasons
Traced manual decisions: agent, timestamp, reason, documents consulted
Roles and permissions per team, full access log
Conversion, rejection and fraud statistics by country and by product
Orchestration

Your decision rules, country by country and product by product

A savings account opened remotely and a re-registration in store do not call for the same level of assurance. Naya ID lets you write those differences down: thresholds, required documents and the cases sent to review are configured per country, product and risk level, with no redeployment of your application.

Match thresholds and assurance levels per product
Required documents and fallback flows by country of residence
Automatic routing to human review on the reasons you define
Rule versioning: every decision stays tied to the rule in force
POST /v1/verifications
{
  "document": "id_card",
  "country": "SN",
  "policy": "savings_account_v3",
  "callback_url": "https://your-app.com/hook"
}

→ 201  { "id": "vrf_8Kd2", "status": "pending" }
→ hook { "status": "approved", "duration_s": 43,
         "policy": "savings_account_v3",
         "evidence_url": "/v1/verifications/vrf_8Kd2/evidence" }
Implementation

Four ways to plug Naya ID in

The same decision engine, whichever deployment mode you settle on with your technical teams and your regulator.

REST API
Three calls is all it takes: create the verification, receive the decision through a signed webhook, retrieve the evidence file. A test environment is available from sign-up.
Mobile and web SDKs
A ready-made capture flow in your brand colours, in French, English and Portuguese. Mobile: Flutter (iOS and Android). Web: JavaScript, usable with any framework, and Angular.
Hosted link
A single-use secure link sent by SMS or email to verify a customer with no development, useful in branches, agent networks, or to bring an existing portfolio into compliance.
On-premise deployment
For the most regulated players, the platform runs inside your environment: no outbound data, compliance with residency requirements, guided deployment.
Frequently asked

What technical and compliance teams ask us

How long does a full verification take?

The median observed time from capture to decision is 43 seconds, AML screening included. Cases routed to human review depend on your teams’ availability; the console shows waiting times and the age of queued cases.

What happens when a document is not recognised?

The flow offers a fallback defined with you: a new guided capture, an alternative document accepted for that country, or routing to human review with the images and scores obtained. No customer is left without a way forward, and the failure is logged to feed document coverage.

Can you collect documents other than identity documents?

Yes. Depending on your rules, the flow can request complementary documents: vehicle registration and insurance certificate for an insurer, proof of address, business paperwork. They are read by OCR and added to the evidence file. Native anti-fraud authentication remains reserved for the identity documents of the 12 covered countries.

Which SDKs do you offer, and for which platforms?

One mobile SDK and one web SDK, running the same flow. On mobile, the Flutter SDK covers iOS and Android and is published on pub.dev. On the web, the JavaScript SDK is added with a script tag or through npm, with no dependencies and whatever your framework; an Angular version is also published on npm. If your customer opens the flow on a desktop, the web SDK offers to hand off to their phone by QR code for the capture. With no app or website, the hosted link and the REST API remain available.

Do the SDKs need a stable connection?

Capture happens locally and quality checks run on the device. If the connection drops, the flow resumes at the last validated step, without asking the customer for documents already submitted.

How are your decision rules audited?

Every rule version is dated and retained. A decision stays tied to the version in force when it was processed, which makes it possible to explain, months later, why a case was approved or refused.

Can we switch on only one block?

Yes. Many projects start with document verification and biometrics, then add AML screening and the review console as volumes and obligations grow. The blocks share the same evidence file.

Next step

Let’s run the flow on your documents and your countries