Know a fridge is failing before the vaccines inside it are ruined.

The problem

A fridge can fail quietly for hours before anyone notices.

Rural clinics depend on continuous refrigeration to keep vaccines and temperature-sensitive medicines usable. A compressor fault, a door left ajar, or a power outage can push a unit out of its safe range long before a staff member happens to check the thermometer — and by then, the stock inside may already need to be discarded. Imithi ColdGuard is designed to catch that drift while there is still time to act, not after the fact.

The solution

A monitoring layer built for clinics with limited staff and limited time.

Shown here against a simulated monitoring programme across four real South African district municipalities — Vhembe (Limpopo), OR Tambo (Eastern Cape), Zululand (KwaZulu-Natal) and Pixley ka Seme (Northern Cape).

Simulated network, real locations

Architecture

Multi-model AI with human confirmation at every decision.

No AI recommendation discards stock or shuts down equipment on its own. Every alert carries evidence, confidence and requires human confirmation. Throughput figures are simulated target design metrics, not measurements from a live deployment.

Illustrative scenarios

Common cold-chain failure patterns, and how Imithi ColdGuard responds.

Illustrative industry scenarios based on well-documented cold-chain failure categories, anchored to real South African district municipalities for context — not reports of real incidents or real clinics.

A compressor fails gradually, not suddenly

Scenario 01 · Slow temperature drift · Vhembe District

The situation

A fridge compressor starts under-performing. Instead of failing outright, internal temperature climbs by half a degree every hour — from 5°C to 9.4°C over six hours. No alarm sounds because the unit is still running; it is just running less effectively, and a manual spot-check earlier that day showed a normal reading.

How Imithi ColdGuard responds

Azure ML tracks the trend line, not just the current reading, and flags the sustained upward drift once it crosses a safety-relevant threshold — well before a manual check would likely catch it.

< 60s

Drift crosses threshold to alert

~3s

Maintenance alert generated

Hours → minutes

Time to first detection

AI models: Azure ML — trend detection · GPT — plain-language alert

Door left ajar overnight

Scenario 02 · Door-open event · OR Tambo District

The situation

After a busy immunisation day, a fridge door does not fully latch. Temperature rises slowly overnight while the unit runs continuously against the open seal. By morning the contents may already be compromised if the event is not noticed early.

How Imithi ColdGuard responds

Door-open duration is tracked alongside temperature. An extended open state triggers an alert even before the safe range is breached, giving staff time to close the door and check stock.

AI models: Sensor rules · GPT — prioritised alert wording

Intermittent power outage

Scenario 03 · Power loss · Zululand District

The situation

Grid power drops for 40 minutes. The fridge stays cold initially but begins to warm. Staff may not be on site or may assume the unit will recover without checking residual cold mass or recovery time.

How Imithi ColdGuard responds

Power-state sensors fire immediately. Combined with temperature trajectory, the system estimates time-to-breach and notifies the responsible contact with clear next steps.

AI models: Power + temp fusion · Claude — history of similar outages at site

Founder

“We built Imithi ColdGuard so that a rural clinic does not have to discover a failed fridge by opening the door and finding ruined stock.”

— Imithi team · product intent (illustrative)

Platform

What operators see

A single view of units across a district cluster: live temperature, power, door status, trend classification, and an audit trail of AI explanations and human confirmations. Demo data only — not a live deployment.

Company

Built for South African primary-care cold chains

Imithi focuses on the practical constraints of rural and peri-urban clinics: intermittent connectivity, limited technical staff on site, and the need for every automated recommendation to stay under human control.

Contact

Talk to the team

Imithi ColdGuard is a product of JujuTech. For partnership or pilot discussions about cold-chain monitoring in primary care settings, reach out to us directly.

Demo content only. District names are used for geographic context in simulated scenarios and do not imply any official relationship with those municipalities or the Department of Health.