Start with a single business process that meets these criteria:
- High volume – it happens often enough that improvements will have real impact.
- Pain point – staff, clients or customers often experience frustration with it.
- Data-rich – it involves information that AI could potentially process.
- Contained – it has clear start and end points.
Examples of processes:
|
Retail
- customer returns
- inventory stocking
- customer enquiries
|
Consultancy
- client onboarding
- proposal creation
- invoicing
|
Hospitality
- booking management
- customer feedback
- roster scheduling
|
Healthcare
- appointment scheduling
- patient intake
- referral processing
|
When you’ve chosen a process, work out how much time, effort and resourcing it takes from start to end. Include details for each step of the process, in the order it happens. For example:
- action – what happens at the start of the step
- who – the person, role or system responsible
- inputs – what information is needed to progress
- outputs – what the result or outcome is
- time – how long the step usually takes to complete.
Why this matters
When you include the people responsible for a process, it helps ensure any AI you adopt works the way you need. It also highlights who could be affected if the process changes.
You should keep an up-to-date record of all AI systems you use. This helps you keep people informed and to know the risks.