Define
Describe the task, intended users and information involved.
AI Tools
A custom AI tool is most useful when it supports a clear task. We plan purpose-built tools around the people who will use them, the information they need and the limits of the chosen technology.
Make the tool fit the task
We clarify what the tool should help people do, then assess feasibility and requirements before development. Data access, third-party AI services, privacy considerations and ongoing support are discussed and agreed for each project.
Scope
Agree essential features, technical dependencies and limitations.
Build and check
Develop the agreed tool and review it against the intended task.
Design for the task and the people using it
A custom AI tool needs a clear input, an expected output and a way to recognise when that output is not reliable. We plan the interface and checks around how the work is actually done.
Define before building
- Who will use the tool and for what task?
- What information can it access?
- What output is useful and how should it be reviewed?
- What should happen when confidence is low or data is missing?
Testing and ongoing needs
Test with representative examples, including edge cases, and keep a person involved where incorrect output could matter. Agree data retention, access, external service costs and update responsibility before launch. AI output can be inaccurate and should be checked for its intended use.
Thinking about a tool for a specific task?
Tell us what you want the tool to help people do.