CHALLENGES

Problems you might be facing

You know AI is relevant but not where to start

The technology moves fast and the noise around it is significant. Working out which applications are genuinely useful, as opposed to impressive in a demo, takes time most teams don't have.

Generic tools solve generic problems

Off the shelf AI products are built for the broadest possible use case. When your process is specific, a general purpose tool needs so much configuration and compromise that you've done the work anyway.

Previous attempts haven't delivered anything usable

Early AI projects often produced prototypes that worked in controlled conditions and fell apart in practice. Unreliable output, staff who didn't trust it, no clear owner. The technology gets blamed for the approach.

The business case is hard to make in the abstract

Convincing stakeholders to invest in AI is difficult when the benefit is theoretical. What tends to work is a small focused application solving a visible problem, which builds confidence for what comes next.

Featured solution

Natural language search for a town council

A Cornish town council approached us about website search because WordPress websites are not effective when it comes to the way people search and the way content is structured.

The council gave us a summary of logged telephone enquiries grouped by topic. Refuse and recycling, public toilets, fly tipping, parking. A large share were not town council business and needed passing to the county council. Many others were simply requests to reach a named member of staff.

We built a prototype AI assistant that interprets a question as written and gives direct advice, rather than returning a list of pages to read through.

Outcome

  • Questions answered in plain English, not a list of page links
  • Enquiries for other authorities identified and routed
  • Existing website content reused without a rewrite
  • Short answers with a link to the source page

Built from the council's own enquiry data. Available to demonstrate.