Where should your business start with AI?

There’s no shortage of AI tools to choose from. The harder question is where one would make a worthwhile difference to your business.

Start with a task your team already knows too well: reading the same kinds of enquiries, copying information between systems, preparing routine documents or chasing missing details.

Your first project should solve a recognisable problem, have a clear owner and produce a result you can measure.

Find the work that keeps repeating

Speak to the people doing the work. Ask them:

• What do you repeatedly copy, write, check or chase?
• Where does work sit waiting for someone to pick it up?
• Which tasks take longer because information is missing or scattered?
• What still depends on someone remembering to do it?

Look beyond the most frustrating task. Something that takes ten minutes several times a day may be a stronger starting point than an awkward job that happens once a quarter.

Write down a few candidates, including how often each happens, who handles it and roughly how much time it takes.

Choose a task with a clear beginning and end

“Automate our admin” is too broad to build or evaluate.

“When an enquiry arrives by email, capture the customer’s details, prepare a brief and assign it to the right person” gives you something concrete.

For each candidate, establish:

• The trigger: what starts the work?
• The information: what does the task need, and where is it stored?
• The result: what should be ready when the task is finished?
• The exceptions: when should a person step in?

If your team cannot agree on the correct outcome, clarify the process before automating it. Otherwise, you risk making an inconsistent process run faster.

Work out where AI actually helps

Some tasks follow straightforward rules. Moving a form response into a customer database may only need a standard automation.

AI becomes relevant when the task involves interpreting information: understanding an enquiry, extracting details from varied documents or preparing a draft from notes.

A useful system might combine both. AI reads the enquiry and prepares the brief; automation saves it to the right place and notifies the team.

The choice should follow the task. There is little value in adding complexity where a simpler solution works.

Start with an output your team can check

Consider this illustrative first project.

A business receives enquiries containing different combinations of requirements, locations, deadlines and attachments. Someone reads each message, picks out the important details and forwards it to a colleague.

An initial setup could prepare a structured brief containing:

• The customer’s contact details.
• What they need and where.
• Any deadline they mentioned.
• Missing information that needs following up.
• A suggested person or team to handle it.

A member of staff checks the brief before it is used. If a detail is missing, the system flags it rather than filling the gap with a guess.

This keeps the initial scope manageable and gives the team a clear way to judge the output. Sending replies or making bookings can be considered once the first stage works reliably.

Measure the benefit honestly

Before building, record a baseline using a representative sample of real tasks. Include time spent checking, correcting and chasing information.

Then compare the same measures during the trial.

Also check whether work is more accurate, responses are faster and fewer tasks are left unfinished. Time freed creates capacity; it does not automatically become a cash saving. Weigh the benefit against setup costs, running costs and ongoing oversight.

Expand once the first project earns it

Give one person responsibility for the trial. Agree what success looks like, which actions need approval and what happens when the system cannot complete a task.

Use suitable information, control who can access it and check how any connected tools handle it.

Run the system through ordinary cases and awkward ones. A useful first project should be dependable enough for the team to use in their normal working day.

Once it proves worthwhile, you have a stronger basis for connecting the next step. That is how a focused improvement can grow into a wider business system.