Exception handling is the job
The happy path is already automated. What consumes your operations team is the delayed shipment, the split delivery and the customs query.
Logistics & supply chain
The plan is fine. It is the exceptions that cost you.
Every logistics operation has a system that handles the normal case beautifully and a group of experienced people who handle everything else by phone. That group is the real system. They are also the constraint, the single point of failure, and the reason the business cannot grow without hiring proportionally. Encoding what they know is the highest-return software in the sector.
What buyers here open with
“Can you handle our exceptions, not just our happy path?”
30%
Conversion lift when we replaced a distributor’s CRM with their real process
The pressure
Four things that decide whether volume is profitable.
Partner data quality
Carriers, suppliers and customers all send data in their own shape, on their own schedule, with their own definition of on time.
Visibility gaps
Customers expect to know where their order is. Most of the chain was never instrumented to answer that.
Margin per movement
Small per-unit margins mean the operational cost of handling an exception can exceed the profit on the order it belongs to.
The work
What we build for logistics operations
Four systems, all of them aimed at the work your most experienced people currently do by phone.
Order and fulfilment orchestration
Routing an order through the right path, including the paths that only apply on Tuesdays to one customer.
Exception workbenches
The screen your operations team lives in, ranking what needs attention by cost rather than by age.
Carrier and partner integration
One internal model, many external formats, with the translation and the failure handling in one place.
Track and trace
A customer-facing answer to a simple question, assembled from a supply chain that was not built to give one.
What an exception costs
The five that consume your operations team, priced.
These figures are the shape we see across logistics clients rather than a quote for your business. The point is the ratio: handling cost against the margin on the order it belongs to.
Address exceptionBad or incomplete delivery address. Caught late, it becomes a failed delivery and a return leg.
£22per occurrence
Stock shortfallPromised inventory that is not physically there. Costs a call, a decision and usually a discount.
£9per occurrence
Customs queryDocumentation mismatch on a cross-border movement. Long tail, high touch, specialist knowledge.
£78per occurrence
Failed deliveryNobody there, refused, or damaged. Redelivery plus the customer contact that follows.
£41per occurrence
Split shipmentOne order, two movements. Doubles the tracking surface and confuses the customer.
£16per occurrence
Margin per orderFor comparison. One customs query consumes the margin on two orders and change.
£34contribution
Non-negotiable
Four rules that keep exceptions visible
- Model the exception
- If your data model only describes the happy path, every exception becomes a free-text note and the knowledge stays in someone’s head.
- Idempotent partner feeds
- The same shipment update will arrive twice. Processing it twice must not create two shipments.
- Time zones and cut-offs
- Almost every logistics bug is eventually a time bug. Store instants, render locally, and be explicit about cut-off boundaries.
- Degrade to visible
- When an integration breaks, the work must surface to a human immediately rather than sit in a queue nobody is watching.
Where AI actually lands
Four places it earns its cost, and one line we do not cross.
Worth doing
- Exception triage, ranking what an operator should handle next by expected cost rather than arrival order.
- Document handling for customs, bills of lading and proofs of delivery, which is high volume and well specified.
- ETA prediction from your own historic performance rather than the carrier’s optimistic estimate.
- Demand and capacity forecasting, where a small improvement moves a lot of working capital.
Where we stop
Autonomous commitments to customers. A predicted delivery date presented as a promise is a customer-service problem you have automated the creation of.
Evidence
A specialist distributor had configured a well-known CRM for two years and still ran the real pipeline in a spreadsheet, because the CRM described someone else’s process.
Read the case study
How we deliver it
The services this work draws on.
Custom Software Development
Systems built for one business, owned outright, with no licence renewal attached.
AI Modelling & Development
Production ML pipelines, retrieval systems, and evaluation harnesses built to survive audit.
Enterprise Software Development
Platforms that hold up under real load, real compliance, and real headcount.
Other sectors we work in
Next step
Talk to the engineer who would run your build.
No discovery call with a salesperson, no deck. A senior engineer reads what you send and replies with a real assessment, including when we think you shouldn’t build it.
