NairaMeter combines smart metering with AI trained to flag theft, tampering, and billing anomalies across your network — with a plain-language reason behind every alert, not just a score.
A trained model scores incoming meter data for patterns consistent with known theft and tampering signatures — surfacing what warrants a closer look, rather than every anomaly at once.
Voltage, current, power factor, and tamper status are captured on an ongoing basis across every connected meter, forming the baseline the model measures against.
Each flagged case is presented with the specific factors that contributed to it, so a technical or commercial reviewer can evaluate it on its merits.
Reviewed and confirmed cases can be incorporated into future model retraining, so detection is designed to improve with continued use.
Each partner's data is accessed under credentials scoped specifically to that partner's own records.
Detection is only the first step. NairaMeter is built around a complete review loop, so a flagged case leads somewhere — and every outcome, right or wrong, makes the platform sharper.
The model flags a reading pattern consistent with theft or tampering, with a confidence level and the specific factors behind it — never a bare score with no explanation.
A designated reviewer on your team examines the flagged case directly on the dashboard, alongside the underlying consumption, voltage, and power data that triggered it.
Each case moves through a clear status — open, under investigation, confirmed, or false positive — visible to your whole team, not just whoever first reviewed it.
Once a case is confirmed, an estimated revenue impact is calculated and reflected in your portfolio-wide totals — or held as "not yet estimated" where the underlying data doesn't yet support a figure, rather than guessing.
If a confirmed case later turns out to be a mistake, it can be reversed — removed from confirmed totals, with the correction recorded rather than the case simply deleted.
Confirmed cases become part of what the detection model learns from in future retraining — so accuracy is designed to improve the more the platform is actually used, not stay fixed at launch.
All data moving between the dashboard, our servers, and our database is encrypted using industry-standard TLS, at every point in the chain.
Each partner is issued a dedicated access credential, enforced on our servers to restrict visibility to that partner's own data exclusively — not merely hidden by the interface, but refused at the source regardless of what is requested.
Every request for data is individually authenticated; there is no unauthenticated access to partner information at any point.
Data shared with us remains the property of the partner that shared it, and is not used, shared, or disclosed beyond the agreed scope of the engagement.
We continue to invest in additional safeguards, including formal third-party security review, as we scale to larger deployments.
NairaMeter is designed to accommodate a range of integration approaches — including scheduled file-based exports and automated data-ingestion pipelines — depending on what your existing metering infrastructure supports.
At minimum: voltage, current, power factor, active/reactive power, and recorded consumption, at whatever interval your meters already report — commonly daily or sub-daily readings.
Once a data-sharing arrangement is in place, incoming data can be validated and integrated automatically on a recurring schedule, reducing ongoing manual handling for your team.
Engagements typically begin with a small, defined pilot covering a limited number of meters, to confirm data compatibility before any broader integration is scoped.
Incorporated in Nigeria as NairaMeter Limited (RC 9810900).
A live platform, actively processing real meter data from partner networks.
Detection model validated at 99.4% ROC-AUC on internal test data, ahead of full field deployment.
In active discussion with independent power producers, mini-grid operators, and regulatory stakeholders across Nigeria's electricity sector.
Embedded and industrial-scale power supply arrangements, where a single flagged anomaly can represent significant recovered revenue.
Community and rural distribution networks, where estimated billing and unrecorded consumption are persistent operating challenges.
Structured, service-based metering partnerships, including models that don't require upfront meter procurement.