What Are The Three Things You Hate Doing In Your Business?

Quick Takeaways

1. Most AI advice for small business starts with a tool and then goes hunting for a problem to attach it to, which is why so much of it lands as either hype or a threat.
2. A better place to begin is the question I open every workshop with: tell me three things you hate doing in your business right now.
3. The answers are remarkably consistent across trades, clinics, shops and agencies. Quoting, chasing invoices, answering the same question forty times a week, writing anything at all, and following up the people who went quiet.
4. Against those five jobs, today's AI is genuinely strong on first drafts, on answering repeat questions, and on never forgetting to follow up.
5. It is weak anywhere a wrong answer is expensive. Numbers, prices, promises, and any job that needs to know what you personally would have said.
6. Time each of your three for a week, put an hourly value on the total, and you have a business case that owes nothing to anybody's sales pitch.
7. Then solve one. The first thing that actually works will teach you more than a year of reading about it.

The demo was the problem

I spent the first stretch of running AI sessions for business owners doing it completely backwards. I would turn up with a laptop, put something clever on the screen, and watch the room go polite. Polite is the worst thing that can happen in a workshop. Nobody argues with you, nobody asks a real question, everyone writes the name of the tool in a notebook and never opens it again. At the time I read that as a failure to explain the technology properly, so I explained it harder, with more screens, which made it considerably worse.

What fixed it was giving up on the demo and asking a question before I showed anyone anything. Now I sit down, and before the laptop is even open I ask the same thing every time: tell me three things you hate doing in your business right now.

The temperature in the room changes inside a minute. People who arrived with their arms folded start talking over the top of each other. Somebody always says quoting. Somebody always says the phone. There is usually one person who admits, with a slightly guilty look, that they have not chased an unpaid invoice since about June. Ten minutes in I have a whiteboard covered in real problems belonging to real businesses, and every single one of them is a better starting point than anything I could have put on a screen.

Why starting with the technology gets you nowhere

Almost all the AI advice aimed at small business begins with a product and works backwards to find a job for it. Here is the tool, here are its features, now go and imagine where it fits. That order of operations creates two responses and neither of them is useful to you.

The first is hype. Everything changes, the future is here, you are already behind. The second is threat, which is the same message with the lighting turned down. Both of them ask you to take a position on a technology in the abstract, which is an impossible thing to do while you are also quoting a job, ordering stock and working out why the ute is making that noise again. So you do the sensible thing and file it under later.

Starting from a job you already resent does something different. It gives you a problem with edges. You know how often it happens, roughly how long it takes, how much it costs you when it goes wrong, and precisely how much you dislike doing it at half past eight on a Sunday night. That is a testable brief. You can try something against it on a Tuesday and know by Friday whether it helped, which is a far more honest experiment than anybody's launch video.

There is a second reason the question works. The jobs you hate are usually the jobs you are worst at, because you have been avoiding them for years. The quote that sat in drafts for four days, the invoice you felt awkward chasing. Those gaps cost more than the time it would take to close them, so the emotional answer and the commercial answer point the same way for once.

The five that come up nearly every time

I have asked this question in rooms full of builders, physios, retailers, accountants and people running online shops, and the same handful of answers keep arriving.

Quoting

Quoting is the most common answer by a distance, and the pain has two halves. There is the writing, which is mostly the same paragraphs reshuffled for the hundredth time. Then there is the pricing, which is judgement built on years of knowing what a job like that actually turns into once the wall comes off.

AI handles the first half well. Feed it the scope, your previous quotes and your terms, and it will produce a structured, readable document in under a minute that sounds like your business if you have set it up properly. The second half stays with you. Anything that hands over your numbers without you looking at them is a mistake waiting for a bad week to happen, and I would not let a machine anywhere near a price until you have watched its drafts for a month.

Chasing money

Nobody enjoys asking for money they have already earned. Most owners delay it, then send something slightly too apologetic or slightly too sharp because they have worked themselves up about it.

This is one of the cleanest wins available, because the hard part here is admin and admin is the one thing a machine never finds awkward. A polite reminder on day seven, a firmer one on day fourteen, a phone call flagged at day twenty one. You want the escalation written by you once and triggered automatically after that, with a human eye on anything involving a client you actually care about keeping.

The same question forty times a week

Do you service my suburb, what are your hours, do you take EFTPOS, how much roughly, can you come Tuesday. Across a week this quietly eats hours, and it arrives through five different channels while you are up a ladder.

A well-built assistant handles this better than most owners expect, as long as it has been given real answers and a hard boundary at the edge of what it knows. The businesses that regret it are the ones that let it guess. We build our Billy Bots with a short list of things they are allowed to say and an instruction to hand over to a person the moment a question steps outside it, because a confident wrong answer about a price costs you more than a slow right one.

Writing anything

Website copy, the newsletter you have not sent since March, the social post, the follow-up email. People hate this one out of proportion to how long it takes, because staring at an empty page is a particular kind of misery.

AI is very good at removing the blank page and moderately bad at everything after that. What comes back is a competent, slightly beige draft that reads like it could belong to any business in your category. Your job is the surgery afterwards. Cut the adjectives, put in the detail only you know, and make sure that something in there could only have been written by your business.

The people who went quiet

Every business has a list of people who asked for a price and were never heard from again. Most owners know this and still do nothing about it, because following up feels like admitting you want the work.

Machines have no feelings about that at all. A system that flags every quote at day three and day ten, drafts a short message and puts it in front of you to send, will find you money that was already sitting in your inbox. This is usually the highest return item on the whole whiteboard and almost nobody names it first.

Where it still falls over

I would rather you hear this from me than find it out on a live client. AI is unreliable with numbers, and it is unreliable in a way that looks calm and confident, which is the dangerous combination. It will produce a total that is plausible and wrong, and it will do it in the same tone it uses when it is right.

It has no idea what you would have said. It can copy your sentence structure after a while. Knowing that this particular customer is a bit anxious and needs a phone call before the invoice is yours alone.

It also inherits whatever you feed it. Point an assistant at a website with old pricing and last year's service area and it will repeat that beautifully to every person who asks. Automating a broken process gives you a faster broken process and a bigger mess to unpick.

There is also a maintenance cost the demos skip. Something you set up in March needs checking in July, because your prices moved and nobody told it. Budget an hour a month per thing you automate.

The twenty minute exercise

Do this before you buy anything. A kitchen table, a piece of paper and twenty minutes is the entire toolkit.

1. Write down the three things you hate doing. First instinct, no editing, no trying to sound strategic about it. If four come out, write four.
2. Estimate how long each one took you last week. Actual hours across the whole week, including the ten minute versions that happen six times. Most people undercount this by about half.
3. Put a number on your hour. Divide what the business turned over last year by the hours you personally worked. It will be an uncomfortable figure. Use it anyway, because it is closer to the truth than the rate you would quote a client.
4. Multiply, then times it by forty eight. Now you have the annual cost of each job in dollars. This is the moment the exercise stops feeling like a workshop activity.
5. Circle the one with the highest number and the lowest risk if it goes slightly wrong. Chasing quotes is usually a safer first project than pricing them. Start where a mistake is survivable.
6. Give it four weeks and one success measure. Hours back, quotes out the door, or replies from people who had gone quiet. Write the measure down before you start, because otherwise you will remember it generously.
7. Only then go looking at tools. With a defined job, a cost and a measure in hand, you will be able to tell inside ten minutes whether something actually fits.

What to do next

The question works because it starts where you already are. Your three answers are specific, they are yours, and they are measurable, which is everything the abstract conversation about artificial intelligence is missing.

This is exactly how our AI Fluency for Business Owners workshops open, before anybody touches a tool. Four sessions, small cohorts, and the first hour is spent on whiteboards covered in things people are sick of doing. If you want to sit with the question somewhere a bit less formal first, there is a free community called The Billy Bot Blueprint on Skool where this kind of thing gets pulled apart in public, mistakes included. It is at skool.com/the-billy-bot-blueprint-4837 and it costs nothing to sit in the back and read.

And if you would rather just talk it through, that is what we do at Infinity1 most days of the week. Bring your three. We will tell you which one is worth solving and which one you should keep doing yourself for now.

Frequently asked questions

What if my three things are all completely different? That is normal and it does not change the exercise. Costing all three is what tells you which one to attack first.

Is my business too small for any of this? Solo operators often get the most out of it, because there is nobody else to hand the admin to. A one person business that stops losing two hours a week to repeat questions has gained a real afternoon.

Should I fix my website before automating anything? Usually yes. Anything you build reads from your site, and so do the AI search engines that customers now ask about you. We score sites out of 100 for exactly this reason, and we will not run ads for anyone until that number is above 75.

How long before I know whether it worked? Four weeks on one job, measured against the number you wrote down at the start. If nothing has moved by then, the process underneath probably needs fixing first.