Rural solar water infrastructure

Don't wait for
the pump to fail.

IoT sensors and AI-powered predictive maintenance flag early signs of solar borehole pump failure — before a community loses access to water.

Sensor detects anomaly → AI predicts risk → technician is dispatched before failure.
98 pumps healthy
Watch: rising vibration
PP-00124 · Demo Village Operational
Health
92%
Failure risk
Low
Solar output
87%
Last telemetry
2 min ago
Vibration trend — 14 days
Watch: vibration trending upward over 8 days. No action required yet — model is monitoring.
How it works

From sensor to technician, before the pump stops.

Sense → connect → predict → alert → repair. Every stage keeps a human in control of the final decision.

01

Sense

Vibration, current, voltage and solar sensors continuously monitor pump behaviour.

02

Connect

Telemetry transmits securely through AWS IoT Core, even over poor connectivity.

03

Predict

Azure Machine Learning analyses patterns and flags anomalies against normal behaviour.

04

Alert

Pump Pulse raises a maintenance risk — never a guarantee of failure, always a probability.

05

Repair

AI drafts a work order and routes it to the right qualified technician nearby.

Why it matters

Protect infrastructure. Protect water access.

Predict before failure

Identify abnormal pump behaviour early, while there's still time to act.

Reduce downtime

Give technicians advance warning instead of an emergency call-out.

Protect infrastructure

Extend the useful life of donor-funded and public water infrastructure.

Protect water access

Help communities keep reliable access to the water source nearest to them.

Predictive maintenance

A probability, not a prophecy.

Pump Pulse never tells an operator a pump will fail on a specific day — it shows the evidence and an estimated window.

  • 1

    Trend detected

    Vibration, current or solar output drifts outside the normal operating range.

  • 2

    Model estimates risk

    A health score and failure-risk band are calculated from historical patterns.

  • 3

    Confidence is shown

    Every prediction ships with a confidence percentage — never framed as certainty.

  • 4

    Work order drafted

    AI generates the recommended action, required skills and equipment.

  • 5

    Human decides

    A technician or manager can always override the recommendation, with a reason logged.

WORK ORDER · PP-00124HIGH PRIORITY
Issue

Increasing motor vibration over 8 days.

Likely cause

Possible bearing degradation — not yet confirmed.

Signals contributing to prediction
Vibration trend ↑
Motor current variance
Operating temp ↑
Prediction confidence
82%
Estimated maintenance window

7–14 days — inspection recommended, not an emergency.

Under the hood

One platform, purpose-built providers for each job.

Pump Pulse is not locked to a single cloud or a single AI vendor. A routing layer sends each task to whichever provider handles it best.

AWS IoT Core

The connectivity backbone. Every sensor authenticates with a unique certificate and publishes telemetry — vibration, current, voltage, solar output, flow, water level — into a secure, time-stamped pipeline built to tolerate patchy rural connectivity.

Azure Machine Learning

Runs the anomaly-detection and failure-prediction models — turning raw telemetry trends into a health score, a failure-risk band, and the specific signals behind each prediction.

Azure OpenAI + GPT models

Drafts work orders, technician-facing explanations and operational summaries in plain language, and powers the technician assistant that answers diagnostic questions in the field.

Claude models

Reviews ambiguous or conflicting sensor patterns and drafts the more detailed maintenance reports and root-cause write-ups — the tasks that benefit from careful, nuanced reasoning.

Google Cloud AI (Gemini)

Reads technician-submitted photographs — a cracked panel, corrosion, loose wiring — and flags a potential issue for human confirmation. Never a substitute for a physical inspection.

GPT Image 2

Generates illustrative system diagrams, sensor-placement guides and training visuals — always labelled clearly, and never presented as real sensor evidence.

Data flow

Sensor to decision, in one pipeline.

Sensors
AWS IoT Core
Time-series storage
Azure ML
AI orchestration
Work order
Technician
Model feedback
Fail-safe by design: pumps keep operating normally if cloud connectivity drops. No AI model can remotely disable a pump — only a human can act on its recommendation.
Dashboards

The same telemetry, framed for each role.

A resident sees a status. A technician sees a work order. An NGO sees a portfolio.

Illustrative demo data — not live infrastructure
For residents

Residents get a status, not a spreadsheet.

No sensor jargon, no false alarms — just whether the water is working, and a heads-up if maintenance is planned.

  • 1

    Check water

    See the current status of the nearest pump in seconds.

  • 2

    Report a problem

    Strange noise, low pressure, or visible damage — reports feed straight into the AI model as a supporting signal.

  • 3

    Receive updates

    A short, plain-language notice when maintenance is scheduled — nothing alarming, nothing technical.

Your water system

Status✓ Working normally
Last checkedToday
Water serviceAvailable
Maintenance is scheduled for your water system tomorrow between 09:00 and 12:00. Water will remain available during the visit.
Why it matters

Reliable water. Smarter maintenance. Stronger communities.

Illustrative figures — every deployment's real numbers depend on the pumps, sensors and communities it monitors.

124
pumps monitored
38
communities supported
27
preventive interventions
19
failures caught early
DEMO DATA
Pricing

Priced in Rand, scaled to your infrastructure.

Actual monitoring cost depends on sensors, connectivity and AI usage per pump — figures below are illustrative estimates.

NGO programme
R180–R360/pump/month
Illustrative estimate
  • Per-pump monitoring
  • Predictive maintenance alerts
  • Programme-level dashboard
  • Technician work orders
Talk to us
Municipal programme
Portfolio pricing
Configured per deployment
  • Everything in NGO programme
  • Infrastructure-wide dashboard
  • Maintenance backlog & response-time tracking
  • Technician capacity planning
Request a proposal
Government & enterprise
Custom deployment
Volume-based, ZAR
  • Large-scale infrastructure monitoring
  • Dedicated onboarding
  • Full audit logging & RBAC
  • SLA-backed support
Contact sales
Questions

Frequently asked questions

Does the AI ever shut off a pump? +

No. The AI can only raise a recommendation. Pumps keep operating normally, and only a human technician or manager can act on an alert.

What happens if a sensor loses connectivity? +

Telemetry buffers locally and syncs once the connection returns. An offline sensor is shown as "offline," never automatically treated as a pump failure.

How confident is a failure prediction? +

Every prediction ships with a confidence percentage and the specific signals behind it. Pump Pulse always frames outcomes as a probability, never a certainty.

Can a technician disagree with the AI? +

Yes. Every recommendation can be overridden, with the technician's reason logged and fed back into future model evaluation.

Is community and technician data kept private? +

Yes. Access follows role-based permissions, infrastructure locations aren't exposed publicly, and all changes are recorded in an audit log.

Know before the pump stops

Start monitoring your first pump.

See pump health, predicted risk and technician dispatch working together — before a community goes without water.

Monitor a pump
From our director

"Asikhathali ukuthi indawo isekude kangakanani — inhloso yethu ukuqinisekisa ukuthi amanzi ayahamba, njalo. Pump Pulse was built so that no rural community wakes up to a dry tap and a pump nobody knew was failing. We watch the pumps so you don't have to find out the hard way."

Lethu Mpilo Tshabalala
Founder & Director, Pump Pulse — a product of Ekasi Gloss Works (Pty) Ltd
SIMULATED TELEMETRY — for demonstration only
Live monitoring

PP-00124 · Demo Village

A running view of what a pump's live telemetry feed looks like — sensor readings, the data pipeline, and the event log that feeds the prediction model.

System view

Streaming
Solar panel Controller Pump Borehole TANK Water tank AWS IoT Illustrative system diagram — not to scale

Vibration & current — last 60 seconds

updating every 2s
Vibration (mm/s) Motor current (A)

Telemetry event log

Live

Register your organisation

Set up access for your NGO, municipality, technician team or facility.

NGO / funder

Monitor a programme of pumps

Municipality

Manage regional infrastructure

Technician

Receive and complete work orders

Platform admin

Configure sensors and users

Enter your first name.
Enter your last name.
Enter your organisation's name.
Enter a valid email address.
Enter a password
Password must be at least 8 characters.
Passwords don't match.

Account created

Check your inbox to verify your work email, then log in.

Go to login