Where to Start With AI in Your Business (Non-Technical)

If you run a business, you’ve probably felt it: a low, constant pressure that you should be doing something with AI, mixed with no clear idea where to actually begin. Every week there’s a new tool, a new “you’re falling behind” headline, and a pile of advice that contradicts the advice from last week.

Here’s the good news. You don’t need to become an AI expert, and you don’t need to “adopt AI” across your whole company. You need to find one process where AI clearly recovers hours for you, and start there. This guide walks you through how to spot that first opportunity without wasting money on the wrong tools.

Stop asking “how do I use AI?” Start asking “where does my time leak?”

Most people start with the tool. They see a demo, get excited, and go looking for a place to use it. That’s backwards, and it’s why so many AI experiments quietly get abandoned.

Flip the question. Instead of “how can I use AI?” ask “where is my team’s time leaking out of the business?” AI is just a way to plug specific leaks. If you don’t know where the leaks are, no tool will help.

Spend twenty minutes writing down the tasks that eat your week. Don’t filter yet, just list them. Write down the small ones too, the two-minute jobs that repeat all day. Those are the ones that never make it onto anybody’s list, precisely because each one feels too trivial to mention. Add them up before you decide they don’t matter.

A useful prompt if you get stuck: what did you or your team do last week that felt like typing something a computer already knew?

The four signs a task is a good first candidate

Once you have your list, score each task against these four signals. The best place to start is the task that hits the most of them.

1. It’s repetitive. You or someone on your team does it the same way, over and over, pulling the same report, sending the same kind of follow-up, copying data from one place to another. Repetition is what machines are good at. If the task changes shape every time it comes up, it’s a poor first candidate no matter how annoying it is.

2. It follows rules, not judgment. “If a new lead comes in, send this email and add them to that sheet” is rule-based. “Decide whether to fire a client” is judgment. Start with the rule-based stuff. A good test: could you write the task down as a set of instructions clear enough for a competent new hire to follow on their first day without asking you anything? If yes, it’s rule-based enough to start with.

3. It’s a bottleneck. When this task backs up, other things wait on it. Automating a bottleneck frees more than just the minutes, it unclogs everything downstream. If a quote sits for days before it goes out, that isn’t a scheduling problem, it’s a lost-deal problem, and it costs more than the twenty minutes the quote itself took to write.

4. Mistakes cost you. Manual steps get skipped or done inconsistently, especially when people are busy. If a missed step here means a lost lead, an angry customer, or a compliance headache, consistency alone is worth a lot. This one is easy to undercount, because the cost shows up somewhere other than the task itself, often weeks later and attributed to something else.

A task that’s repetitive, rule-based, a bottleneck, and costly when it slips? That’s your starting point. You’ve just found it without touching a single tool.

The trap: automating the wrong thing first

The most common failure here isn’t picking a bad tool, it’s picking a bad first project. It’s tempting to automate the thing that’s most exciting, or most visible, instead of the thing that quietly costs the most.

Two traps in particular:

• The shiny trap. You automate something flashy that demos well but barely moves the needle. It feels like progress; it isn’t.
• The everything trap. You try to overhaul five processes at once, get overwhelmed, and stall out entirely.

Pick one boring, expensive, repetitive process. Boring is good. Boring is where the real hours are hiding.

There’s a third trap worth naming, because it’s the one most likely to cost you your team’s goodwill. It’s automating a process that is genuinely broken. If a step in your workflow doesn’t work when a person does it slowly and carefully, automating it will not fix it. It will simply produce the wrong result faster and more consistently, and now nobody trusts the system. Fix the process on paper first. Then automate the version that works.

A simple way to size the opportunity

Before you build anything, do quick napkin math so you know whether it’s worth it:

(Minutes the task takes) × (Times per week) × (Weeks per year) ÷ 60 = Hours you’d get back

A 15-minute task done 10 times a week, across 52 weeks, is 130 hours a year, more than three full work weeks. That’s the number that tells you whether a process is worth automating, and it’s the number that makes the decision obvious instead of emotional.

If the hours-back number is small, move to the next task on your list. If it’s big, you’ve found something real.

One caution on the math. Count the whole task, not the part you picture. A report that takes ten minutes to assemble usually also involves remembering it’s Friday, finding last week’s version, chasing one number somebody hasn’t sent yet, and fixing the formatting. The remembering and the chasing are part of the job, and they can easily outweigh the task itself. Time the task honestly once, start to finish, before you trust the total.

What “starting” actually looks like

Here’s the part nobody tells non-technical owners: your first AI project should be small and contained. You’re not rebuilding your business. You’re taking one process and making it run with less manual effort.

That might look like:

• New inquiries automatically sorted, logged, and answered with a first response, so nothing sits unread in an inbox.
• A weekly report that assembles itself instead of someone copying numbers every Friday.
• Client onboarding steps that fire in the right order automatically, so nothing gets skipped.

None of these require you to understand the technology underneath. They require you to know your process, which you already do, and to connect the right tools to it.

Notice what these have in common. Each one has an obvious before and after, each one either works or visibly doesn’t, and none of them puts the business at risk if it needs a week of adjustment. That’s what a good first project looks like. If you can’t explain to your team in one sentence what will be different on Monday, the project is too big or too vague to be first.

Give it a few weeks before you judge it, too. Expect a round of tuning once real work starts flowing through it, because real work tends to contain cases nobody thought to mention. Budget for that from the start, rather than reading it as a sign the thing was a mistake.

You don’t have to figure this out alone

Here’s the honest trade-off. You can learn the tools yourself. There are AI assistants, automation platforms like n8n, and integrations that can do all of the above. If you enjoy tinkering, that’s a real path, and plenty of owners take it.

But most business owners don’t want a new hobby, they want the time back. Trying to learn the tooling while running your business is often where the “where do I start” question turns into “I never got around to it.” The tools are not hard, exactly. They’re just one more thing, and one more thing is what you already have too many of.

That’s the whole reason this kind of help exists. The value isn’t knowing which button to click. It’s knowing which process to start with, and building it so it actually holds up in the real world. That’s the thinking behind the audit and the build work I do, and it’s why the first conversation starts with your process rather than with tools.

Where this leaves you

If you take one thing from this: the first move isn’t choosing a tool, it’s finding the leak. Write down what eats your week, score each item against the four signals, do the napkin math on the winner, and be honest about whether the process works before you automate it. That’s a genuinely useful afternoon, and none of it costs anything.

More notes on this collect under Worth Automating if you want the next one. You can find that first opportunity yourself, without any help from me. The reason to bring somebody in is speed and the cost of a wrong first pick, not because the thinking is beyond you.

Ready to find your first opportunity?

If you’ve got a process in mind, or you just know your team is losing hours somewhere and can’t pin down where, Start a Conversation. Tell me what’s taking up too much time, and if there’s a strong fit, I’ll follow up to help you spot the clearest first win. You can also read how the work actually runs, or more about me if you’d rather know who you’d be talking to first.