Anyone can now ask a chatbot how to write a survey report, plan a menu or file a tax return. So do your years of experience still matter? More than ever. Here's why, and how to find the part of your work that's worth turning into software.
What AI is good at
AI tools are very good at the general stuff. They have read the textbook, the guidance notes and a million forum posts. Ask them a standard question and you'll get a decent standard answer, fast. They're also good at the boring writing: turning notes into a tidy paragraph, summarising a long document, drafting an email.
That's genuinely useful. But it's also exactly what everyone else has access to as well.
What you have that AI doesn't
If you've done the job for ten or twenty years, you carry things no chatbot has:
- Judgement. You know when the textbook answer is wrong. The surveyor who has seen a hundred 1930s houses knows which cracks matter and which don't. The accountant knows which "simple" question hides a problem.
- Context. You know how the work really happens on a wet Tuesday, not how the manual says it happens. You know the shortcuts, the workarounds and where things go wrong.
- The real problems. You know which tasks waste hours every week, because you do them. Most software fails because it solves a problem nobody actually has. You don't have that risk.
- Trust and access. People in your industry listen to you. You know the customers, the suppliers and the people who'd pay for a better way. That's the hardest thing for any new product to get.
- Standards. You know what "good" looks like. A chatbot can produce a report. You can tell whether it's a report a client would accept.
Why this matters more now, not less
AI has made building software faster and cheaper. That sounds like bad news for experts, but it's the opposite. When building gets easier, the scarce thing is no longer the code. It's knowing what to build, and for whom.
That's you. The most valuable software of the next few years won't come from people who know how to code. It'll come from people who know a problem inside out, working with people who can build.
And the best products will use AI inside them, guided by an expert's rules. A tool that drafts the survey report in seconds is handy. A tool that drafts it and rates every defect the way a twenty-year surveyor would is a business.
How to find the process worth turning into software
You probably already have a few ideas. Here's how to find the strongest one.
1. Keep a "sigh list" for a week
Every time a task makes you sigh, write it down. Note three things: how often you do it, how long it takes, and what goes wrong. Don't judge the list yet. Just collect.
2. Look for these five signs
- It repeats. Weekly or daily, not once a year.
- It follows steps, even if one or two of them need your judgement.
- It hurts. It costs time, money, sleep or risk. A failed inspection, a late invoice, a missed deadline.
- Others do it too. Other people in your industry do the same task the same painful way.
- It's held together with tape. Spreadsheets, copy-and-paste, photos in a WhatsApp group, a Word template you've edited for years.
A task with all five is a strong candidate. A task with only one or two is probably just annoying.
3. Find where your judgement comes in
Look at the steps and ask: where do I make the call? That's the heart of your product. It's the bit a generic tool can't copy, and the bit you can build in as rules, checklists or ratings.
4. Test it with three questions
- Can I describe the problem in two sentences to someone outside my industry?
- Do I know at least five people who'd recognise it straight away?
- Would they pay to never do it this way again?
If the answer to all three is yes, you have something worth exploring.
An example
A building surveyor takes around 200 photos on a survey, then spends four hours at night typing up the report. The steps are the same every time. Their judgement comes in when rating each defect. Every surveyor in the country has the same problem.
The software: take photos room by room on site, rate each defect using the surveyor's own guide, and the report writes itself. The surveyor checks it and sends it. Four hours becomes a few minutes, and the product can be sold to other surveyors.
What to do next
Pick the strongest item from your list and write it down in a short paragraph: what goes wrong, how often, who else has the problem, and what a better way would look like. That paragraph is enough to start a conversation.