Here’s how to start using AI in your business without the overwhelm: pick one repetitive task, work out what it actually involves, and hand the boring part over.
You’re sitting in a business coaching group, watching your coach use Claude in a way you’d never have thought of. Her work takes half the time it should, and it lands with her clients looking polished.
Or someone casually mentions the automation they built in Zapier, the one that follows up with every new enquiry while they’re asleep.
Or you join a webinar about turning a discovery call into a project proposal with Google Workspace, and you think: I want some of that.
Then you close the laptop and the actual business lands back on you. Invoices. A client waiting on files. That thing you promised yourself you’d sort out in September.
And the question that follows is never “should I?” It’s “where on earth do I even start?”
There are so many tools. So many options. Half of them sound identical, and you’ve got no reliable way of telling which one would genuinely help you and which is noise built for a business nothing like yours.
So here’s what I’d suggest instead of a big tech overhaul. Pick one repetitive thing. Work out what it actually involves, step by step. Then see whether a tool you already pay for can take part of it off your hands.
That’s the whole method. It’s much less glamorous than the webinar, and it works.
Why does everyone else seem to have this figured out?
They mostly don’t.
I talk to founders all the time, and this comes up more than anything else: tech and tool overwhelm, and no idea where to begin. They know there’s something out there that would save them hours of mundane work. They just can’t find the energy to learn a new skill on top of everything else, which I think is completely reasonable.
Learning new technology has a real cost. Time, attention, and that slightly humbling fortnight where you’re rubbish at something. When your week is already full, that cost looks enormous sitting next to a vague promise of saving time later.
So make the cost small. Small enough to fit in one Tuesday afternoon.
Where should you start?
With the task you groan at.
Not the most complicated part of your business, and not the thing you think you “should” fix. The one that makes you sigh every single time it appears, because it’s the same shape every single time.
Monthly invoicing. The onboarding email you’ve written 40 times. Sorting receipts. The report you rebuild from scratch each month. Whatever it is for you, start there, because repetitive and predictable is exactly what these tools are good at.
How do you map a process you’ve never written down?
Most founders have never actually written this stuff down, which is why it feels slippery. So before you look at a single tool, map the process out as if you were handing it to a new team member.
→ Write the steps in order, from the thing that kicks it off to the moment it’s done. You can use dictation option in Claude or ChatGPT, or Wispr Flow, and ask your AI tool to turn it into something cohesive.
→ Note what you need in front of you at each step. A contract, a spreadsheet, last month’s figures, a folder of files.
→ Be specific about the documents. Do you write a contract from scratch every time, or do you have a template you fill in? (If it’s from scratch, you’ve found your first fix before you’ve even opened a new tool.)
→ If a client sends you things regularly, note how they send them, and what you do the moment they land.
→ If you produce a report, write down what goes into it and what makes it genuinely useful for the person reading it.
This bit is dull. I know. But it’s also where most of the time savings turn up, because half of what you’re about to automate shouldn’t exist in that form anyway.
Which tool should you use?
Start with the ones you’re already paying for. You’d be surprised how much is sitting unused behind a subscription you renew without thinking.
If you pay for Notion, have you tried getting Notion AI to draft a project proposal for you, based on the meeting notes it took?
If you use Claude or ChatGPT, have you asked it to write itself a set of instructions for drafting your emails, in the tone your clients expect?
Have you tried pulling your expenses out of a banking statement with it? (Hide or remove anything confidential first, please.)
If you’re still typing up notes during discovery calls, there’s transcription software that does it for you while you concentrate on the human in front of you. Otter and Granola both do the job.
If nothing you own can help, have the conversation with your AI tool. Describe the process, tell it how the finished result needs to look, give it an example if you can, and ask what would do the job within your budget. Then take the shortlist and start a free trial. If it’s not what you hoped for, try the next one. Nobody gets this right the first time.
Is the setup time actually worth it?
I’d think about it this way: how much time does the manual version eat every week?
If the answer is 20 minutes a month, leave it alone. Genuinely. Go and do something nicer.
If the answer is an hour a day, like it is for founders piecing together their to-do list from a calendar, an inbox and two task lists, then a couple of hours of setup pays for itself before the month is out.
I won’t lie to you. The effort is front-loaded on purpose. It’s not the most comfortable thing to go through, but I believe the pay-off is very much worth it. You put in a bit extra for a short stretch, and you get hours back on a continuous basis afterwards. The goal was never to add another project to your plate (I promise).
What does this look like in a real business?
I’m in the middle of one of these right now, building an AI agent for a client and her team of five.
At the moment, every monthly report on a social media account’s performance comes out slightly different. The broad shape is the same, but each person focuses on different things, presents them in a different format, and the design is all over the place, because everyone is doing what they think is best. Which is a perfectly sensible thing for people to do when nobody has told them otherwise.
So I’m building templates the agent refers to every time it produces a particular report. Same structure, same standards, whoever on the team sets it running.
It isn’t finished yet, so I can’t tell you what it saved them. What I can tell you is that most of the work so far has been deciding what a good report looks like, not building anything clever. That’s usually how it goes.
What do I actually use myself?
Fair question, and the honest answer is: I automate a great deal for my clients and surprisingly little for myself. Builder’s own house, and all that jazz.
The one thing I’d struggle without is Granola, which sits in my meetings and writes up the notes. Useful on its own. What makes it genuinely valuable is what I do with those notes afterwards.
When a new client comes through the door, I build the scope of the project straight from the notes of that first conversation, instead of staring at a blank page trying to remember what they said about their invoicing.
And when I’m partway through building a system for someone, I go back through the transcripts to work out where the build is actually heading. The detail that changes everything is almost always something they mentioned in passing, five weeks ago, while talking about something else.
One tool, doing two jobs that used to be mine. That’s the size of thing I’m suggesting you start with.
What’s a “skill”, and why would you want one?
You’ll hear the word “skill” a lot at the moment. It means a set of written instructions you give an AI tool so it does a task your way, every time, without you re-explaining yourself.
The useful part: these are just files. Most of the AI tools run on much the same infrastructure, so instructions you write for Claude will usually travel over to ChatGPT, Gemini or Perplexity with very little fuss. You’re not locking yourself into anything.
Take client onboarding. You’d walk the tool through the process you already follow, step by step, and it writes the instructions for itself. (With Claude you can record a video of yourself doing the thing, and it’ll turn that into written instructions, which feels mildly like witchcraft the first time.)
A rough onboarding skill would cover:
→ Your contract template
→ Your payment terms
→ Your invoice template
→ The actions you always take when a new client comes through the door
→ The things every client has to give you before you can start
Write it once, and it works the same way on the Friday you’re exhausted as it does on the Monday you’re sharp.
What should you never hand over?
Anything that relies on your expertise, your judgement or your taste. That’s the secret sauce that makes your work yours, and no tool is going to reproduce it.
Hand over the parts that don’t change: the formatting, the chasing, the copying from one place to another, the first rough draft you were going to rewrite anyway.
Do you actually have to automate anything?
Honestly? That’s up to you. You’re the decision maker in your own business.
If writing your to-do list by hand each morning is how you ease into the day, keep it. That’s a ritual that grounds you in the work, and handing it to a robot would be a downgrade.
But if you’d rather an agent assembled that list for you, pulling from your calendar, your inbox and three different task lists, because the manual version eats an hour before you’ve done anything useful, then that process deserves a proper look.
Frequently asked questions
How long does it take to set up my first automation? For one small, well-understood process, an afternoon is usually enough, and most of that goes on mapping rather than building.
Do I need to pay for more tools? Start with what you’ve already got. Most founders are paying for more capability than they use.
What if I pick the wrong tool? You probably will at least once. Use free trials, keep the first attempt small, and move on without ceremony if it isn’t right.
I’m not technical. Is this still for me? Yes. The hard part is describing your own process clearly, and you already know it better than anyone.
The bottom line
Technology has got good enough to genuinely take the boring parts off your hands, and it’s worth using that to your advantage.
Not so you can cram more in. So you can work calmly, keep the business sustainable, and take an occasional Thursday afternoon off without the nagging feeling you’re falling behind.
Pick one task you dread doing. Map it out. See what a tool you already own can do with it. That’s the whole first step.
If the tools are here to help you do what you’re best at with more joy, why not use them?
If you’d like more of this sort of thing, I write about operations for growing businesses over on my Substack. Come and subscribe -> Link to Substack
And if the real problem is that you can’t see where the repetitive work is hiding in the first place, that’s an operations question rather than a tool question. It also happens to be exactly what I do (hello! 🙋♀️), so book a chat and we’ll have a look together -> link to a Free Ops Audit
About the author
Jagoda Puczko is a business operations partner who helps growing founders close their laptop at 6pm. She has run operations for teams of 100+ at Apple and Little Dot Studios, directed award-winning films, and once had a headshot photography business with a very sophisticated pricing strategy.
Leave a Reply