When a custom application makes sense
Consider a custom application when a specific workflow or integration cannot reasonably be supported by existing tools. First validate the need and responsibility for operation. A feature list alone is not a reason to build.
Before commissioning development
- Who will use the application and which task will they complete?
- Why is a process change or an available product insufficient?
- Which data does it need and who may access it?
- How will it connect to existing systems and who manages changes to them?
- Who funds and manages operation, support, updates and a possible exit?
When comparing options and costs, include staff effort, maintenance, integrations and decommissioning. Base the budget on validated requirements and proposals.
What the brief should contain
| Area | What needs to be clear |
|---|---|
| Work scenarios | Starting situation, user, steps and expected result |
| Data and access | Sources, permissions, organisation isolation, retention and deletion |
| Traceability | Which changes and approvals to record and who can inspect them |
| Integrations | Inputs, outputs, formats and behaviour when another service is unavailable |
| Acceptance | Test scenarios, acceptance criteria, open defects and required documentation |
| Operation | Owner, monitoring, backups, recovery, updates and support |
| Exit | Data export, documentation, code licensing and migration support |
Choose technology for these requirements and the skills of the team maintaining it. Neither cloud nor self-hosting guarantees security on its own.
Test one complete scenario first
The pilot should demonstrate a useful task from input to output. For an AI governance application, this could mean creating a record, adding evidence, review and approval by an authorised person, and exporting the result. Test rejection, missing data and unauthorised access too.
Before wider deployment, test backup recovery and handover of administration. Decide whether to continue based on results, cost and known limitations.
How this relates to AI Governance Studio
This guide helps you prepare a brief for a custom tool. The current AI Governance Studio and assessment process are presented on the service page with demonstrations. Features and their use during an engagement depend on the specific offer.