Prove that what was built is working, and fix it fast when it is not.
Solar sensors on school roofs, walls and water tanks send tilt, water, power and photo evidence. An edge model flags changes, GPT-class models draft alerts in local languages, an image model checks cracks and leaks, and a named officer approves every response. Municipal and traditional-authority dashboards see the same record.
This solution requires current AI models to operate.
Verify delivery
Is the roof dry, the tank filling, the power on? Measured, not assumed.
Local languages
Alerts reach principals, parents and leaders in the language they use.
Leaders included
Traditional authorities and municipal officers share one view.
Why delivery gaps repeat
- 1
Funds are allocated against a long backlog - 2
Work is done or promised on paper - 3
Nobody measures the result - 4
Roofs, walls, tanks and wiring decay - 5
The school reports it, but it is not scheduled - 6
Response comes late or to the wrong place - 7
Learners lose days; the building becomes unsafe - 8
No record, so the cycle restarts
Three kinds of case
Slow decay
Wall movement or a slow leak that shows up over weeks.
Service failure
The tank runs dry or the power goes off.
Claim versus reality
Work is reported complete but the sensors disagree.
Decision support only. Responsibility stays with the humans and authorities who approve action.
How it works
- Ingest. Tilt, water level, power and photos arrive from each site.
- Score. A model estimates days until a limit is crossed.
- Explain. A GPT-class model writes the reason in plain words.
- Pattern. A ranking model compares schools and seasons.
- Verify. An image model checks a photo of the crack, leak or tank.
- Visualise. Dashboards show status to officers and leaders.
- Human decides. A named person approves, changes, rejects, orders a field check or escalates.
AI helpers and cloud helpers
Chosen from this product’s own steps. Verify every model and service name against provider documentation before launch. None are partners or endorsers.
Structure watcher
Tilt and vibration readings → movement flag. Runs at the edge (for example NVIDIA Jetson with TensorRT).
Water and power forecaster
Tank level and battery history → days to empty. Cloud (for example AWS SageMaker).
Crack and leak verifier
Photo → confirmed damage or not. Image model (for example Amazon Rekognition or a Vertex AI vision model).
Delivery checker
Reported repair vs. sensor change → “fixed” or “not yet”. Compares claims with measurements.
Alert writer
Evidence → short alert in 11 languages. GPT-class model (OpenAI or Azure OpenAI); a fluent reviewer checks it.
Backlog ranker
Many schools’ risk scores → who is visited first. Long-context model (for example Gemini on Vertex AI).
Human reviewer
Decides everything. Nothing auto-approves.
Cloud helpers
Device ingest (Azure IoT Hub or AWS IoT Core), storage, event routing, WhatsApp and SMS messaging, audit log.
This solution requires current AI models to operate; the models listed (or their current equivalents) are essential to the production system.
3D model: the school
3D is not available in this browser. The system has a classroom block, a water tank, a solar roof, sensors, a gateway and a tablet.
How this model works. The sensors on the wall and the water tank take measurements all day. The solar roof powers them. The gateway sends the numbers on. The tablet shows ALL OK, WARNING or ALERT. When you slide to “unsafe”, the wall leans a little, a crack appears, the water tank drains and the tablet changes. A person then decides what to do.
Evidence
Five sources supplied with the brief, on service gaps, school backlogs and housing rights. Open each and check it before launch. Pilot measurements, not these articles, will show how well the system works.
| Source | What it shows | What the system does about it |
|---|---|---|
| Underdeveloped infrastructure and service gaps in townships and rural areas gov.za notice, June 2026 upload | Government recognises service barriers. | Gives officers measured delivery data. |
| Gwarube faces a R140-billion school infrastructure crisis allAfrica, September 2026 | National backlog; rural closures and disparities. | Backlog ranker puts the riskiest schools first. |
| UFH study on Eastern Cape school infrastructure gaps University of Fort Hare | Provincial backlog; unsafe structures and absenteeism. | Structure watcher flags movement early. |
| SA Human Rights Commission report on schools TimesLIVE, 12 November 2025 | Unsafe, under-resourced schools across provinces. | Audit trail records who decided what. |
| Flooded and Forgotten: informal settlements and the right to housing ReliefWeb | Poor services, including access to education, are rights failures. | Local-language alerts reach affected communities. |
About CHARLIE TECH
“A school should never have to wait for a wall to fall before someone listens. We build the tools that let communities see problems early, prove what is true, and act while there is still time.”Simphiwe Thanyani Mulaudzi, Founder and Director, CHARLIE TECH
CHARLIE TECH is a South African technology company registered in Gauteng. We build the Yield Verification Integrity and Response System so that officers, principals and traditional leaders share one measured record of what has been built and whether it is working.
Company registration
- Enterprise name
- CHARLIE TECH
- Registration no.
- 2026 / 896684 / 07
- Enterprise type
- Private Company
- Status
- In Business
- Registered
- 7 November 2026
- Financial year end
- August
- Registered office
- Hamerkop Street, City of Tshwane Metropolitan Municipality, Gauteng, 0157
Download registration certificate (PDF)
Source: CIPC COR14.3 Registration Certificate issued 7 November 2026.
Leadership
- Director
- Simphiwe Thanyani Mulaudzi
- Role
- Founder and Director
- Appointed
- 7 November 2026
- thanyani@charlietech.co.za
Contact
Direct email
Founderthanyani@charlietech.co.za General enquiriesinfo@charlietech.co.za Administrationadmin@charlietech.co.za Requests for quotationrfq@charlietech.co.zaCHARLIE TECH · Reg. no. 2026 / 896684 / 07
Registered office: Hamerkop Street, City of Tshwane, Gauteng, 0157
Legal and privacy
In short: a live product in pilot stage. The Contact form opens your own email app; we store nothing here. No cookies, no trackers, no selling of personal information. People, not software, decide in the Workspace. You can ask to see, correct or delete your information.
Who we are
CHARLIE TECH, a Private Company registered in South Africa (registration no. 2026 / 896684 / 07). Registered office: Hamerkop Street, City of Tshwane Metropolitan Municipality, Gauteng, 0157. Director: Simphiwe Thanyani Mulaudzi. Privacy and general queries: info@charlietech.co.za.
This site
A public product site and the Open Pilot Workspace, which uses pilot test data. A name typed at sign-in stays in browser memory and clears on reload or Reset.
Personal information
Contact form details are used only to answer your enquiry, on your consent, and kept only as long as needed (period to be set by the company).
Security
HTTPS in transit, access controls, data minimisation and retention limits. If a breach affects personal information we will notify the Information Regulator and affected people as the law requires. No internet transmission is perfectly secure.
Third parties
Google Fonts, the Three.js CDN (cdnjs) and Open-Meteo load when you visit; your own email provider handles messages you send. Processing may occur outside South Africa.
Your rights
Access, correction, deletion, objection, withdrawing consent, and not being subject to solely automated decisions. Use the Contact topic “Data or privacy request”. You may complain to the Information Regulator.
Terms and liability
No scraping, reverse engineering, attacks or load-testing, and do not present pilot outputs as real alerts. Content, designs, models and code belong to the company. Provided as available. Outputs are decision support, not professional advice; responsibility stays with the humans and authorities who approve action. Liability is limited as far as the law permits; nothing excludes liability that cannot lawfully be excluded.
Law
South African law (POPIA) applies for South African users; international visitors are responsible for their own legal compliance. English text governs translations.
Have a South African attorney review this page before launch.
Open Pilot
Pilot testing access. Nothing you type is sent to us.
Schools
Site A · water tank
% tank full
Battery 94% · Signal -74 dBm · Solar roof charging
Decision queue
No recommendation yet. Pick a school, inject a fault and watch the reading move. Nothing is ever approved automatically.
Work orders
Empty until a person approves.
Field kit
Tasks appear after approval.
Audit trail
| Time | Who | Action | Why |
|---|---|---|---|
| No decisions yet. | |||