Business cases

How I turn a business request into a working AI solution.

Every AI engagement I run starts with the same six questions, before any architecture, model choice or tooling is discussed. Most failed AI projects did not fail on the technology. They failed because nobody agreed what problem was being solved, for whom, or how anyone would know it worked. The framework below is the discovery process I apply to every use case, followed by anonymised examples drawn from production work in the property and development sector.

Business problem
What problem are we actually solving?
Target users
Who is this for, and what does their day look like?
Current process
How does it work today, and where does it hurt?
Desired outcomes
What does better look like?
Constraints
What limits, systems, data or rules apply?
Success metrics
How will we know it worked?

The first seven use cases came through a structured AI intake process I designed and ran for a large property investment and development business. Each request was captured against the framework before anything was built, then delivered as an agent on an enterprise AI platform with access to the organisation’s document repositories. The eighth, for an in-house legal team, was built as a self-hosted retrieval pipeline rather than a platform agent.

All cases are anonymised. Organisation names, people, locations, platforms, figures and document contents have been removed or altered, and no confidential material is reproduced. Descriptions are illustrative of the approach, not a record of any specific engagement.