Ask most vendors where to start with AI and you get a list of departments. Sales, HR, finance, support. That list is no help, because it does not tell you which of your own problems AI is any good at.
Here is the version we use with leaders in the room. Every process in your company is held back by something. AI is good at loosening five of those things, and useless against the rest. Find yours, and you have found your first project.
Read the symptoms under each one. Tick the ones that sound like your team.
1. Not enough hands
What it feels like
- You keep hiring and the backlog still grows.
- Your best people spend half their day on work a smart intern could do.
- There is a queue for something and it never gets shorter.
What AI does about it. It does the 500 repetitions so your people do the five that need a decision. This is the most common first build, and the easiest to measure, because you already know how many hours the job takes.
2. Only one person can do it
What it feels like
- Work stops when one person is on leave.
- Everyone waits on the same colleague for the same kind of request.
- A simple report needs someone who can write code, so it never gets made.
What AI does about it. It removes the gate. The person with the question runs the analysis, drafts the clause or builds the small tool themselves, in plain language.
3. The answer exists but nobody can find it
What it feels like
- The same question gets asked in the group chat every month.
- New staff take a quarter to become useful because the knowledge is in people's heads.
- Somebody has definitely answered this before, in a document nobody can locate.
What AI does about it. It becomes the company's memory. Ask in plain words, get the answer with the document it came from attached.
4. Quality depends on who is doing it
What it feels like
- Two people handle the same case and you get two different outcomes.
- Your best work is your best person's work, and it does not spread.
- Checking things is a job on its own.
What AI does about it. One standard, applied the same way every time. The value here is fewer things going wrong rather than hours saved, so measure it that way.
5. Things slip through
What it feels like
- Renewals, follow-ups or approvals get missed and you find out late.
- You would like to treat every customer individually, and there are too many of them.
- Nobody is watching the thing that only matters once a quarter.
What AI does about it. It watches all the time, and it can treat a thousand people individually without a thousand people doing it.
Now count them
Tally your ticks. Your top one or two are where AI pays back fastest. Ignore the rest for now, however interesting they look.
Two things worth saying about the list.
Real work often sits in two of these at once. Personalised follow-up is both a hands problem and a things-slip problem. When that happens the payback is bigger, not smaller.
And speed is not on the list on purpose. Loosening any of the five makes things faster. If you file everything under "we want to be faster", you end up back at a list of tools with no way to choose.
What to do with your answer
Take your top category and name one specific job in it. Not "improve customer support", but "answer the 40 delivery-status emails we get every day". A job that specific can be built, and you will know inside a month whether it worked.
That is also exactly what we do on a first call. Bring the job, and you will get a straight answer on whether AI can do it.
Book a call, or work through the exercise with your own team at our hands-on workshop.