LLM for customer-services automation
We automated call-centre report processing for a UK appliance-repair business - lifting fault-detection accuracy from 50% to 80%.
LLMText classificationOperations
// The challenge
Where it started.
Our client provides repair services for home appliances in the UK and needed to process call-centre reports faster and more accurately.
// Our approach
- 01
Built an LLM-based system to read call-centre reports.
- 02
Detected which part of an appliance is broken from the report text.
- 03
Gave engineers the right part to bring before the first visit.
- 04
Improved prediction accuracy through the model.
// The outcome
What it delivered.
80%
accuracy (from 50%)
1st
visit fix rate
auto
report analysis
Engineers arrive with the right part on the first visit.
Prediction accuracy improved from 50% to 80%.
Faster, cheaper repairs and fewer repeat call-outs.
// More case studies