Clarke Systems Group Request a demo
AI processing and integrations

Manual categorisation, replaced by a measured AI service.

Product categorisation was a manual process. A language-model service now performs it, with the model selected on measured accuracy and cost.

3AI model families benchmarked
AWShosted and access-controlled
MFAwith full audit logging

The problem

Manual categorisation was slow and inconsistent, and the data required strict access control.

The system

A classification service hosted on AWS. Before a model was chosen, OpenAI, Anthropic and self-hosted open-weight models were benchmarked on accuracy against real categories and on cost per classification.

  • model selected on measured accuracy and cost
  • multi-factor authentication and full audit logging
  • a purpose-designed results database
  • integration with existing ticketing and reporting tools
  • continuous accuracy monitoring
Model selection and live classificationIllustrative, not real data

Benchmark on real categories

Model family ASelected
Model family BHigher cost
Model family CLower accuracy

Bars show accuracy. Selection weighs accuracy against cost per item, so the most accurate model is not automatically chosen.

Live classification

Stainless steel 1.7 L kettleStoredSmall appliances › Kettles
Kids' waterproof hiking bootStoredFootwear › Children › Outdoor
Bamboo bathroom accessory setStoredHome › Bathroom › Accessories
The shape of the system: models compared on measured accuracy and cost before one is chosen, then live items classified into the results database, with accuracy monitored continuously.

The result

A manual workflow replaced by a secure service with measured accuracy and a known cost per item, integrated with existing tools.

  • AWS
  • Python
  • SQL
  • LLM APIs
  • REST APIs
  • Dashboards

Designed and delivered by our founder while employed at a global digital-commerce analytics business. Not client work for Clarke Systems Group. Clarke Systems Group is independent and not affiliated with, or endorsed by, any current or former employer.