AI for Logistics in Africa: Tracking, Updates and Exceptions

Karani GeoffreyJune 24, 20266 min read

Spend an hour in the office of an African transport or delivery company and you will hear one question more than any other: where is my cargo? It arrives by phone, by WhatsApp, from clients, from consignees, from the client's client. Between those calls, someone is trying to reach a driver whose phone is off in a network dead zone, someone else is explaining a delay at a weighbridge, and a third person is manually building the day's delivery report in a spreadsheet.

Very little of this is logistics. It is communication overhead generated by logistics. And it is the single largest, most automatable cost centre in the business.

What the day actually consists of

Status enquiries, endlessly

Clients cannot see their shipment, so they ask. Each answer requires someone to find the truck or rider, confirm the position, and translate it into a sentence. Repeat all day.

Manual dispatch coordination

Assigning jobs by phone call, confirming the driver has the address, sending the client's contact, chasing proof of delivery photos afterwards.

Exceptions, which is where the real cost sits

Breakdowns, border and weighbridge delays, wrong or unclear addresses, consignee not answering, failed deliveries, damaged goods, route changes for weather or security. Exceptions are rare per shipment and constant across a fleet.

Documentation and reporting

Delivery notes, proof of delivery, invoicing back-up, daily and weekly client reports. Almost always assembled by hand from photos and messages.

Where the money leaks in African logistics

Coordination labour. People employed largely to relay information between clients and drivers.
Late exception detection. A problem discovered at 4pm that started at 10am costs a whole day.
Failed deliveries from unreachable consignees or bad addresses - the most expensive routine failure in last-mile.
Disputes that cannot be settled because the evidence lives in a driver's WhatsApp thread.
Invoicing delays caused by missing paperwork, which quietly wrecks cash flow.

What AI can handle in a transport or delivery business

Automatic status answers to clients

A client messages asking about a consignment. The AI identifies the shipment, reads the current status from your operational data, and answers immediately with location or stage, expected timing and any known issue. Around the clock, without a dispatcher touching it. This alone typically removes the majority of inbound messages.

Proactive updates before the client asks

Dispatched, in transit, arriving today, delayed, delivered. Sent automatically at the right milestones. Clients who are told what is happening stop calling to find out what is happening.

Structured intake of new jobs

Booking requests arrive as free-form messages. AI extracts the useful fields and creates a proper job record instead of a paragraph someone has to re-type:

{
  "client": "Mwangi Hardware",
  "pickup": "Industrial Area, Nairobi",
  "dropoff": "Nakuru Town",
  "goods": "cement, 120 bags",
  "weight_kg": 6000,
  "requested_date": "2026-08-04",
  "contact_on_delivery": "+254 7XX XXX XXX",
  "special_instructions": "call 30 minutes before arrival",
  "missing_fields": ["exact dropoff address"]
}

Notice the last field. A good extraction step also tells you what the message did not contain, so the AI can ask one precise follow-up question rather than a human discovering the gap at the loading bay.

Driver communication and proof of delivery

Job details sent to the driver, arrival confirmations captured, proof of delivery photos collected and attached to the right shipment automatically. The paperwork assembles itself as the job runs.

Exception detection and routing

This is where AI earns its place. A driver sends tumesimama Salgaa, gari imepata shida ya clutch at 11:40am. The AI classifies it as a mechanical breakdown, links it to the shipment, alerts the operations lead, flags the affected client delivery as at risk, and drafts the client notification for a human to approve. The alternative is finding out at 4pm.

Chasing the information that holds up invoicing

Cash flow in transport is usually blocked by paperwork rather than by clients unwilling to pay. A delivery note without a signature, a missing photo, a job with no recorded completion time - each one delays an invoice by days. AI can check each completed job against what your invoicing process requires, chase the driver or the branch for the missing piece, and flag only the jobs that genuinely need a person. Invoices go out on time because nothing is quietly incomplete.

Reporting that builds itself

Deliveries completed, on-time rate, exceptions by type, vehicle utilisation, which routes and which clients consume disproportionate coordination time. Real operational intelligence out of accumulated history rather than a monthly guess.

What stays human

Route and dispatch decisions in a crisis. Security situations, weather, breakdowns - these need judgement and local knowledge.
Claims for loss or damage. Commercial and sometimes legal consequences. AI gathers evidence, humans decide.
Rate negotiation. Pricing is relationship and margin.
Regulatory and border matters. Customs, permits, compliance - human ownership, always.
Serious client escalations. When a shipment fails badly, the client needs a person who can commit to something.
Driver welfare. Accidents, illness, safety. A human calls, immediately.

Automate the answering, the logging and the alerting. Keep the deciding.

Logistics on the NGAZI ladder

NGAZI, from the Swahili for ladder, describes AI adoption as five rungs where each earns the next and none can be skipped.

Stage	What it looks like in a transport business
Stage 0 - paper and memory	Jobs in a diary, driver updates in WhatsApp, delivery notes in a folder. Nothing countable.
Stage 1 - a fast assistant	Staff use AI to draft client emails and summarise messages. Useful, unconnected.
Stage 2 - it reads your files	AI answers from your real rate cards, SLAs and standard operating procedures, with sources.
Stage 3 - it gets things done	AI answers status queries, creates jobs, sends milestone updates and raises exception tickets.
Stage 4 - it sees the whole picture	AI reasons across years of trips: which routes run late, which clients cost the most to serve, where fuel and time leak.

The honest position for most operators is Stage 0 or 1. If your shipment statuses are not recorded in a system, no AI can answer where is my cargo - because there is nothing to read. Digitising the job record with timestamped status changes is the rung you are actually on.

A realistic rollout

Create a digital job record with a status field that changes at defined milestones: booked, dispatched, in transit, at destination, delivered, exception.
Get drivers updating status through one simple channel. Whatever they will actually use daily beats whatever is theoretically best.
Automate client status answers first. Highest volume, lowest risk, immediately obvious to clients.
Add proactive milestone notifications. Watch inbound enquiries drop.
Then exception classification and alerting. Start with alerting a human, not acting.
Add automated job intake once the record structure is stable.
Review weekly. Every misclassified exception is training material.

Data sovereignty in a competitive market

Your client list, rates, routes and volumes are the most commercially sensitive information in your business. An AI system handling them must keep the data in your environment, with access control and full audit trails, and it should not leave without your authorization. In a market where your competitor may share a client, this is basic hygiene.

Talk to us

upeo.ai is built in Nairobi and designed for operations exactly like this - an AI communication layer across WhatsApp, email and web, connected to live business data, triggering workflow actions such as ticket creation and record updates, while your team stays in control of every consequential decision. Trained on your specific operation, not a generic assistant.

Email: hello@upeo.ai
Phone: +254 116 888 777
Web: https://www.upeo.ai

Start by counting one thing this week: how many inbound messages were simply where is my shipment. That number is your automation business case, and you already have the data to calculate it.

#AI for Business#Logistics#Delivery#Africa#Operations#upeo.ai