Is AI making your company more productive, or just giving everyone more work?

Faster emails and cleaner reports are easy to see. It's harder to tell whether the business is getting any more done.

Imagine a customer asks when their new account will be ready.

Someone uses AI to draft a polished reply in seconds. The reply includes a list of things that need checking, so it gets passed to another person. The colleague opens the request, asks someone else to confirm a detail and sends another message.

The customer is still waiting.

One person's inbox looks tidy. The company's workload has grown.

We're getting very good at making individual tasks faster. The more useful question is whether we're making the business better at getting things finished.

You can count how many emails were drafted, reports produced or meeting notes summarised. Those numbers tell you people are using AI. They don't tell you whether customers are getting a better service, your team is delivering more, or anyone has less to chase.

Start with what you want the business to achieve

Before asking what else you can automate, ask what you want more of.

Perhaps you want to respond to customer enquiries the same day. Get a new member or client onboarded without a string of internal emails. Move quotes through the business faster. Give your team more time with customers instead of filling out forms and chasing approvals.

Once you know the outcome, work backwards. What keeps it from happening today? Where does it slow down? Who has to step in and sort things out?

AI might help. But first you need to see the whole job.

Follow the work, not the tool

Pick one process that happens every week and look at the last few real examples.

Where did the request start? Who picked it up? Which details were missing? How many times was it passed around? Who could approve the next step? And how did anyone know it was finished?

You may find one person has saved time drafting something, while somebody else spends more time checking it. Or a tool produces a perfectly good summary that nobody is responsible for acting on.

Sometimes the tool isn't the problem. The team has never agreed who owns the result. A rule exists only in someone's head. Two systems say different things. Nobody has decided what counts as done.

Adding another AI tool to that process is unlikely to help much.

A founder can end up as the company's human API: the person joining up tools, making decisions and moving work between teams. AI may speed up each task while giving that person even more to coordinate.

Put a number on the chasing

Say a company handles 30 customer enquiries a week. Each one involves, on average, 20 avoidable minutes of checking, chasing or rework across everyone involved.

That's 10 hours a week, or roughly 43 hours a month spent getting work from one person to the next.

If changing the process could remove a quarter of that effort, you'd potentially give the team around 11 hours a month back.

Those numbers are illustrative, not a benchmark or a promise. You'd still need to test whether the work has really disappeared rather than landed in another inbox.

And getting 11 hours back isn't automatically a financial saving. The interesting question is what your team could do with that time instead. Speak to more customers? Get more enquiries resolved? Catch mistakes before they become problems?

A useful starting point is to pick one recurring task, count the checking, chasing and rework across the whole team, and calculate the total time involved.

Change one part of the job, then check the result

The next move doesn't have to be a big AI programme.

It might mean removing a pointless hand-off, agreeing which system holds the correct information, or making it clear who is allowed to approve what.

Where AI can reliably do a repeatable part of the job, give it that part. Keep people involved when judgement or a customer relationship matters. Make sure there's someone responsible for the finished result.

If the goal is faster onboarding, for example, a well-written welcome email isn't the finish line. The customer needs to have completed the right checks and be ready to use the service.

Try the change on a handful of real cases. Check the result in the place where your team normally records it. Look at the time taken, the mistakes, and who still has to intervene.

If it hasn't improved the job, change it again. Or stop.

Make the company capable of more

So if your team is already using AI but the same jobs keep coming back for someone to chase, pick one of those jobs this week and follow it all the way to the customer.

See what actually gets finished, and where the work still gets stuck.

blog