01
Connect your data
Files, APIs or event streams. Start with what you already have.
Financial crime intelligenceFinancial crime intelligence for banks and mobile money
An alert tells you where to look.
RenoTrace tells you what happened.
Turn scattered transaction alerts into evidence you can act on — the accounts, the relationships and the money trail behind a single flagged payment.
Access is reviewed before it is granted · MFA · isolated workspaces
Investigation INV-4C21
GH₵212,500 · 19 accounts
What the evidence shows
- 14 accounts paid in — typical here is 2
- Every payment inside 11 minutes
- 98% withdrawn through one agent
- Three accounts share one handset
AWAITING HUMAN DECISION
An investigator approves every action
Illustration of the product. No customer data is shown.
The problem
Most systems stop at the alert.
Detection tells an analyst that something is unusual. It rarely tells them what happened, and almost never leaves behind something a regulator would accept months later.
A row in an alert queue
→The fourteen accounts that fed it, and the one agent it left through
A score with no explanation
→The specific findings behind the score, written in plain language
A decision nobody can reconstruct
→An evidence trail, the reviewer's name, and the reason recorded
The product
Built for the investigation, not just the alert.
See the whole relationship, not one payment
Wallets, agents, handsets and SIMs connected to the activity under review — laid out so money always reads left to right, and every line is labelled in words rather than symbols.
- Look-alike accounts bundle into one node
- Follow the money past the edge of the picture
- Ask whether two accounts are connected, and how
+9
AI that explains, and never decides
Findings are summarised in plain language with the evidence they came from. Every restrictive action still waits for a named human being to approve it.
- Grounded in the evidence on screen
- No action taken without human approval
- The reasoning is recorded with the decision
Summary of findings
“Fourteen accounts paid this wallet inside eleven minutes. Ninety-eight per cent of the balance left through a single cash-out agent.”
AWAITING HUMAN DECISION
Nothing is blocked, frozen or reported until a person approves it.
Detection that learns from your outcomes
Confirmed frauds, cleared cases and institution-confirmed outcomes feed back as governed training data — so the system gets better at your customers, not at a generic average.
- Labels carry their source and their trust level
- Candidate models run in shadow before they decide anything
- Promotion is gated on evidence, not on a hunch
Investigator reviewtrusted for training
Institution confirmedtrusted for training
External outcometrusted for training
System suspicionnot used for training
A label is only as good as where it came from, so every one carries its source.
How it works
From incoming data to a decision that holds up.
Begin with batch files and grow into real-time scoring when your operation is ready for it.
02
Prioritise real risk
Rules, relationship signals and approved models rank what deserves attention.
03
Follow the money
Expand the picture, trace between accounts, and see the shape of the activity.
04
Record a defensible outcome
The evidence, the decision, the reviewer and the reason — kept together.
Governance
Designed for the day someone asks you to justify it.
In financial crime the question is never only whether the system flagged something. It is whether the institution can explain why, months later, to somebody with authority.
A person decides
The system ranks, explains and evidences. It does not freeze accounts, file reports or close cases on its own.
Every claim traces back
Each finding points at the transactions it came from, so a reviewer can check the reasoning rather than trust it.
New models prove themselves first
A candidate model scores live traffic in shadow, affecting nothing, until there is evidence it is better.
Being wrong is measured
Reviews include an unflagged random sample, so the misses are estimated rather than assumed to be zero.
Security
Trust designed in from the first transaction.
This public site is a static page with no session, no API client and no route into the console. The product itself is a separate, authenticated application on its own hostname.
Reviewed access
Demo and production access is granted deliberately, never self-served.
Role-based permissions
Analysts, reviewers and administrators see and do different things.
Isolated per institution
Each client's data is separated at the database, not only in the application.
Audited by default
Administrative and investigative actions are written to an audit trail.
Give your investigators a clearer view.
Request verified demo access. Each request is reviewed before we create a protected, isolated workspace for you.