← All work
Case study · Logistics

Live delivery predictions across 1,200 trucks.

AI wired into an existing fleet, warehouse and customer-notification stack — predicting delays before they happen.

Client
Atlas Logistics
Sector
Freight
Built in
6 weeks
Live since
Mar 2026
on-time delivery
+31%
on-time delivery
within one quarter
trucks
1,200
trucks
tracked in real time
"where is it?" calls
−48%
"where is it?" calls
to customer support
systems
3
systems
connected, none replaced
The problem

Delays were discovered after they happened.

Atlas had GPS, a warehouse system and a CRM — but none of them talked. Customers learned about delays from an empty driveway, and dispatchers re-planned routes by phone.

The approach

Integrate, don't replace.

We connected the existing systems through an event stream, trained a delay-prediction model on two years of trips, and used an LLM to write clear, proactive customer updates.

  1. Event stream
    GPS · WMS · CRM
  2. Prediction model
    delay risk per stop
  3. Notifier
    LLM-written customer updates
  4. Dispatch view
    re-route suggestions
The outcome

Problems flagged hours earlier.

Dispatchers see at-risk stops hours ahead and re-route before a delay lands. On-time delivery rose 31%, and calls asking where a parcel is have nearly halved.

We kept every system we already paid for. Azed just made them smart enough to work together.

Priya Raman · COO, Atlas Logistics
Next case study · Retail
A cloud POS for small retailers and multi-branch shops
Start a project All work