Pilot scope · Shared solar mini-grids · Off-grid villages, South Africa
Share the power fairly. Show everyone how.
Smart meters on households and generation sources feed AI that spots imbalances and unusual usage, proposes a fair load schedule, and writes reports the whole village can read. The operator and village energy committee make every decision.
Illustrative feeder load · not live telemetry
Illustration: solar supply, a balancer and four metered homes. Original drawing, not a photograph.
Illustration: usage per home against a fair-share line. Original drawing.
What this system does not do: it does not disconnect homes automatically, accuse individuals, set tariffs or make billing decisions. It flags meter patterns as leads. People decide.
01 · The problem
Shared power, shared suspicion
When many homes draw on one solar system, evening peaks overload it, outages follow, and neighbours start to blame each other. Without clear data, conflict replaces cooperation.
Usage differs but stays within the village's agreed share.
Load imbalance
Too many loads at the same hour. Adjust the schedule, not the people.
Possible tampering
A meter pattern does not fit its history. Send a technician to check.
Liability: outputs are leads for human review, never findings of theft or certified determinations.
02 · How it works
From meter reading to fair decision
03 · Intelligence roles
One job per model. A human at the end.
IoT smart meters
Sensors, not AI. Record household use and generation output.
GPT-6 Luna · Anomalies
Detects usage patterns that break from a meter's own history.
GPT-6 Sol · Scheduling
Proposes load schedules and writes fair-usage reports.
GPT-6 Astra · Dashboards
Builds operator dashboards from approved data.
Image 2 · Installation check
Reviews photos of wiring and meters for installation quality.
Human reviewer · Mandatory
The mini-grid operator and village energy committee approve, modify, reject or escalate every recommendation.
Disclosure: model names are as supplied in the project brief and must be checked against current provider documentation. All AI outputs in this prototype are simulated to show workflow shape. Outputs are probabilistic. Monthly cost range: to be confirmed.
04 · Interactive model
Feel the fairness
Balanced
Drag to rotate · scroll or pinch to zoom · simplified for teaching
Blue tilted slabs: solar panels. Balls on houses: smart meters; green is fair, amber is busy, red is unusual. Lines: power shared from the balancer.
Green box: load balancer. Wire shape: the AI thinking. Screen: the fair-usage display. Blue cube: operator. Checks homes when balanced, visits the balancer when busy, goes to the red meter's home when demand is high.
05 · Research & evidence
Why this is needed
Electricity theft and unplanned local outages are reported across South Africa, and shared solar mini-grids are being discussed as an option, along with questions of fairness. These articles were found by search. Open each one and confirm it before launch.
Your brief's five evidence links (a ReliefWeb housing report, a gov.za PDF and four site homepages) do not address micro-grids, so they are not used here.
“I started Mzanzi Logic because too many South African communities are left guessing about who gets power, and why. I want every village to see its energy shared fairly, explained in plain language, with people, not machines, making the final decision.”
DEMO MODE — ALL TELEMETRY, PREDICTIONS AND AI OUTPUTS ON THIS SCREEN ARE SIMULATED
Sign in
Demo access only. No account is created or verified.
Operations board simulated
Meters online
118
Peak load vs capacity
88%
Anomalies awaiting review
2
Work orders issued
0
System status
Battery 71% · Panels 3 of 3 producing · Balancer normal (all simulated)
Recent alerts
18:42 Evening peak near limit 17:05 Meter M-014 pattern unusual Yesterday: Meter offline, cluster D
Household clusters simulated
Human decision
Work orders
Empty. Orders appear only after a human approval in this session.