Manufacturing · 12 months
AI-powered process automation for a manufacturer
A full AI programme across production, from bottleneck analysis to predictive maintenance and MLOps.
- Increase in operational efficiency
- 40%Increase in operational efficiency
- Reduction in production costs
- 30%Reduction in production costs
- Predictive maintenance accuracy
- 95%Predictive maintenance accuracy
The problem
Production processes had grown by accretion, with bottlenecks nobody had measured and maintenance scheduled by calendar rather than condition. Equipment failures were expensive and largely unpredicted.
What I did
Mapped the processes against actual throughput data, built AI tooling for the constraints that were genuinely binding, and put predictive maintenance in place where downtime justified it — then handed over the MLOps to keep it running.
How it ran
- 01
Assessment and planning
Analysed current manufacturing processes, identified bottlenecks, and developed the transformation roadmap.
- 02
Custom tooling
Built specialised AI tools for process optimisation and predictive maintenance.
- 03
Predictive analytics
Deployed prediction models for equipment maintenance and quality control.
- 04
Process automation
Automated key manufacturing decisions where the data supported doing so.
- 05
MLOps and handover
Established MLOps infrastructure and the continuous optimisation process the team now runs.
Other work
All case studies →- Insurance
Sequencing the data foundation before anyone bought an AI agent
A financial group wanted AI scoring across five affiliates. I proposed one affiliate, eighteen weeks, and a data readiness report before a single model — because the alternative is an agent trained on data nobody has reconciled.
- Corporate banking
Winning back the last call after two meetings had failed
A long-standing account was one bad meeting from closing. I rebuilt the pitch around the client's own operating numbers and a working prototype, and cut every claim we could not source.
- Telecommunications
Every clause gets a human verdict
Contract review is the most demoed AI use case in the enterprise and the least trusted. I built the version where the model proposes a redline and a lawyer accepts, edits or rejects each one.
Something similar on your plate?
Thirty minutes, no deck. I will tell you whether it is worth doing at all.