01 / DISCONNECTED
One entity, many records
Names, IDs and attributes vary by source. No single record tells the whole story.
Data infrastructure for fragmented markets
One customer. Many records. Prove the match.
Phanx finds records that belong to the same person or business across disconnected systems, and shows the evidence and confidence behind every match. Built for financial institutions and the developers who build for them, starting in Nigeria.
Early-stage. Built in Lagos.
PHANX · CANDIDATE MATCHIllustrative
System A · Core bankingRecord A
Adaeze OkaforID no. ending 4821 · DOB 12 Mar 1991System B · Lending appRecord B
Adaeze C. OkaforID no. ending 4821 · Phone +234 80x xxx 0142Government ID numberExact match
NameClose variant
Phone numberConflict · flagged
Likely the same person · High confidence
Every signal, weight and conflict is recorded so a reviewer can accept, reject or escalate.
Synthetic example for illustration. Not a product screenshot.
The problem
The same customer shows up as five different people.
In fragmented markets, identity data lives in separate systems with inconsistent names, formats and identifiers. Teams reconcile it by hand, and rarely get to see why two records were treated as one.
02 / UNCERTAIN
Wrong links and missed links
Look-alike records get merged by mistake while genuine matches stay hidden. Both errors cost money.
03 / UNEXPLAINED
An answer isn't enough
Regulators, auditors and reviewers need to see why records were linked, and what is still uncertain.
Our first product
Phanx: entity resolution you can audit.
Most systems return a match score and ask you to trust it. Phanx builds every decision from evidence, so confidence is something you can inspect, not just accept.
01Data
Records from your own systems.
02Evidence
Each comparison becomes a recorded signal.
03Assertion
Signals support or contradict a link.
04Resolution
Weighted, scored, with confidence.
05Entity
One reviewable view of the person or business.
Evidence-based confidence
Scores come from weighted, configurable signals, and conflicts stay visible instead of being averaged away.
Append-only audit trail
Every decision is recorded and can be traced back to the evidence that produced it.
Where we are
Early, and building in the open.
BuiltSchema and resolution engine
Core data model, evidence chain and parametric scoring weights.
In progressReviewer workspace and audit trail
See evidence, resolve conflicts and keep a deterministic record of every decision.
PlannedRelationship graph and API/SDK
Connect resolved entities and give developers a clean integration path.
Built for regulated environments
Your data stays inside your boundary.
Institutions have different infrastructure and governance requirements. Phanx is designed to run within the institution's own environment rather than requiring records to move into a shared platform.
Option 01
Private institutional node
Phanx components run inside the institution's controlled environment. Customers do not have to pool records with anyone else.
Institution-controlledGoverned accessPer-institution deployment
Option 02
API and SDK
For developers and fintech teams who want evidence-backed resolution inside their own products and workflows.
Developer-firstConfigurable weightsEvidence in every response
Where we begin
Nigerian financial institutions.
Banks, lenders and fintechs deal with fragmented identity data every day. It is a demanding place to prove the product. The problem, however, is not unique to Nigeria.
Pilots
Test Phanx on a real problem.
We are looking for a small number of financial institutions and fintech teams to run a scoped pilot with us. We start with sample or synthetic data, and the pilot is shaped around the identity problem you actually have.
1
Pick one problem
For example duplicate customers, KYC reconciliation or merging records across systems.
2
Run it on sample data
We resolve a scoped dataset and show you the evidence behind every match.
3
Decide together
Review results with your team and agree what a controlled production pilot would need.
Team
The people building Phanx.
A geologist who studied fragmentation for two years, a data engineer who builds the systems, and a machine-learning engineer who builds the models.
Adesoye Ibraheem Toyosi
Co-founder
Geologist by training. Spent two years studying why data in fragmented markets will not connect, and set out the evidence-based approach behind Phanx.
Terry Joshua Fregene
Co-founder & CTO
Data engineer and backend developer in Lagos. Builds the modular, Python-based data systems Phanx runs on, and maintains an open-source Python toolkit.
Kofi Opoku
Machine learning engineer
Builds the matching and scoring models. Public work includes a fraud-detection service and applied machine-learning projects.
Why we exist
Read the layers. Show your working.
A geologist reads layers of evidence, each one incomplete and some of them contradictory, to reconstruct what actually happened. Phantera started from that habit. After two years studying why data in fragmented markets will not connect, we saw the same problem inside banks and fintechs: one customer sits in many systems, and nobody can show why two records are the same person. Phanx works the way a careful geologist does. It gathers the evidence, weighs it, keeps conflicts visible, and shows its working.