Back to Portfolio
Madagascar · Mali · Pan-AfricanEnergy · AI · InfrastructureOngoing Engagement180+ Sites Live

AI-Driven Operations for a Pan-African Rural Energy Provider

How we built an end-to-end AI platform that unified data from 180+ solar mini-grids, automated site selection, predicted equipment failures, and powered data-driven growth across sub-Saharan Africa.

180+

Mini-grids managed

30K+

Connections served

99%+

Uptime maintained

900t

CO₂ avoided / year

Scaling Clean Energy Across a Continent

Our client operates solar-powered mini-grids that deliver 24/7 electricity to off-grid communities across Madagascar and Mali, with expansion plans into Nigeria and the DRC. Each site serves 300–400 households and small businesses through prepaid, mobile-money energy coupons and smart meters ranging from 100 W to 16 kW.

As the portfolio grew past 180 sites, the existing toolset could no longer keep pace. Critical decisions — where to build next, how to prevent outages, which tariffs to offer — were being made with fragmented data and manual analysis.

No Unified Visibility

Demand, losses, and outage data were scattered across tools with no real-time, cross-portfolio view.

Manual Site Selection

Satellite and survey data for new village identification was analysed manually — slow and inconsistent.

Demand Forecasting Gaps

Limited ability to forecast energy demand growth or size solar plants efficiently for future load.

Fragmented Customer Tools

Sales, prepaid recharges, and local engagement ran on disconnected systems with no shared intelligence.

An End-to-End AI Operating Platform

Working with the client's operations, engineering, and impact teams, we designed and delivered a five-layer AI solution stack — from a unified data foundation through to demand-stimulation models for community impact.

01

Unified Cloud Data Platform

Consolidated data from smart meters, solar inverters, mobile money platforms, CRM systems, and field surveys into a single cloud warehouse. Standardised historical data from 180+ sites and 30,000+ connections to enable consistent cross-country performance metrics.

Data Engineering
02

AI-Powered Site Identification

Deployed ML models on satellite imagery and geospatial layers — population density, road access, proximity to existing grids — to score village suitability and project demand. Integrated 20,000 previously analysed sites so models could learn from past decisions and outcomes.

Machine Learning
03

Predictive Operations & Asset Optimisation

Built anomaly detection on production and consumption curves to flag irregular losses, potential meter tampering, or early equipment failure. Developed load-forecasting models using weather data, historical usage, and seasonality to right-size new solar plants and plan upgrades.

AI Automation
04

Customer Segmentation & Tariff Optimisation

Segmented customers by consumption profiles, load size, and payment history — distinguishing households, small shops, and productive users like mills and cold rooms. Simulated tariff and coupon structures to balance revenue stability with energy affordability.

Advanced Analytics
05

AI-Supported Demand Stimulation

Built scoring models to identify entrepreneurs most likely to benefit from appliance financing and business support. Linked operational data with impact indicators — new businesses, women entrepreneurs, job creation, CO₂ avoided — to quantify the socio-economic value of each village.

Impact Analytics

From Discovery to Scale in Four Phases

Phase 1Discovery & Alignment

Deep-dive workshops with operations, data, and impact teams covering mini-grid design, 20-year village licences, and local sales processes. Mapped every digital tool in use.

Phase 2Data Foundation & Quick Wins

Rapid consolidation of high-value sources (smart meters, mobile payments) to deliver early dashboards on site performance, losses, and payment behaviour. Anomaly alerts rolled out within weeks.

Phase 3Model Development & Pilot

Piloted AI models in a representative subset of villages in Madagascar and Mali, covering different load profiles and socio-economic contexts. Co-designed operational playbooks.

Phase 4Scale-Up & Continuous Improvement

Progressive deployment across 180+ sites with live financial and impact KPIs. Continuous model retraining as the client expanded to new villages and markets.

Measurable Outcomes at Scale

Operational & Financial Performance

99%+ availability sustained across all mini-grids despite growing portfolio size, supported by AI-driven monitoring

Scalable model proven 180+ electrified villages and over one million people benefiting from access to electricity

Faster, smarter expansion site-scoring models helped prioritise village roll-out and focus capital on highest-impact locations

Customer & Community Outcomes

30,000+ direct connections and hundreds of thousands of indirect beneficiaries with access to clean electricity

5,200+ businesses connected including 1,500+ women entrepreneurs supported through productive-use programmes

~900 tonnes CO₂ avoided annually by replacing diesel alternatives with renewable mini-grids

What We Brought to the Table

Machine LearningSatellite Image AnalysisGeospatial DataCloud Data WarehouseAnomaly DetectionLoad ForecastingCustomer SegmentationPredictive MaintenanceRevenue SimulationImpact AnalyticsReal-time DashboardsMobile Money IntegrationSmart Meter Analytics

Frequently Asked Questions

Facing Similar Challenges?

Whether you're managing distributed infrastructure, scaling across markets, or sitting on untapped data — our AI consulting and development team can help you turn complexity into a competitive advantage.