Most of the return from AI automation is decided before you buy anything. Choose the right workflow, with a number for what it costs you now and someone who owns the result, and the payback usually shows within weeks. Choose badly and you can spend a year on something technically impressive that never shows up in the numbers.
AI is good at amplifying a process that already works. It’s very bad at rescuing one that doesn’t.
What does it mean to put AI into an automation process?
It means the AI sits inside a job your team already does, and the output lands where the work already happens.
A quote gets drafted from your past jobs and a person approves it. An enquiry gets summarised and routed before anyone opens it. A monthly report assembles itself from the systems that already hold the numbers. In each case the AI is a step in an existing process, with a defined input, a defined output and a named human somewhere in the loop.
Most businesses have something else: a subscription, a few people using it privately in different ways and no way of telling whether it made any difference. Both look like “using AI”. Only one compounds.
The practical test: if the person who set it up left tomorrow, would the process still run? If it wouldn’t, you have a tool. A workflow survives the person who built it.
Why start with the business question?
Because the technology question has an infinite number of answers and the business question has about five.
Start with the technology and the conversation becomes which platform, which model, which integration. Those questions come later. Starting with them tends to produce a pilot that works beautifully and never becomes part of how anyone operates.
The business question narrows things fast. Where are we slow? Where do we redo work? What does a customer wait for longest? Which job does everyone quietly hate? Those questions point at specific workflows, and a specific workflow can be scoped, costed and measured.
A buyer’s agent I worked with wanted to take on one more client a year. They preview forty or fifty properties for a single client brief, so the property search looked like the obvious thing to automate. When we looked at where the time went, it was the inbox. The search was the visible work. The email around it was the expensive part, and email rarely gets counted as a workflow.
That’s why the Audit comes first in how we work. Fixed fee, fixed window, and at the end of it you’ve decided which two or three jobs are worth doing.
How do I work out which workflows are worth automating?
Work through them in this order. It takes less time than people expect, and it usually kills a few ideas that sounded good in a meeting.
- Map what actually happens, including the spreadsheet someone keeps privately and the step that only works because one person remembers to do it. The process document won’t show you those.
- Find where it hurts. Look for delays, rework, double entry and the handoffs where things go missing. These are usually well known internally and rarely written down.
- Check the data. If the information the process depends on is scattered, out of date or sitting in someone’s inbox, AI will inherit that problem and produce confident nonsense from it. Fix the input or pick a different workflow.
- Decide what you’re measuring. Turnaround time, error rate, response speed, hours spent. Pick something you can already measure today, so you have a before and after.
- Pick the boring one. The highest-value first project is usually a repetitive, high-volume, low-judgement task. Interesting projects make good demonstrations. Boring ones pay for themselves.
Most businesses find their first candidate in quoting, enquiry handling, reporting or document preparation, because those jobs are frequent, structured and slow.
What does a good implementation involve?
Six things, none of them about the model.
- Start where there’s already structure. If a job has a template, a checklist or a standard format, AI will do well at it. If every instance is bespoke and undocumented, sort that out first.
- Put the output where the work happens. If someone has to remember to go somewhere else, copy something and paste it back, the process will quietly stop being used within a month.
- Name an owner. One person, per process, responsible for the output being right. Including what happens when it isn’t.
- Decide what gets checked. Some outputs need review every time, some need spot checks and some need none. In an automated job that decision becomes a step in the process: a named person checks a set thing at a set point, before the output goes anywhere it can do damage. Leave it undecided and you get one of two failures: output waved through unread, or everything checked twice.
- Choose tools on fit. The right platform is usually the one that connects to what you already run, and we build in the tools you already pay for. If you’re choosing an AI consulting agency, ask whether they resell or earn margin on any platform they’d recommend.
- Put a date on it. Agree when you expect to see a result and what it’ll look like, so the pilot has an end.
More on how this runs in practice on putting AI to work.
What does governance mean for a business this size?
Four written answers: what customer information can go near these tools, who’s accountable for the output, what happens when something comes out wrong and who can change the setup. Teams use these tools cautiously when nobody’s told them what’s allowed, and cautious use produces no return.
If you’re covered by the Privacy Act, the privacy-policy rule on automated decisions that starts on 10 December 2026 can reach a workflow that uses personal information to recommend a decision that could significantly affect a person’s rights or interests, even when someone signs it off. So record what each workflow does as you build it, and go back over the ones already running.
Why does the team decide whether this works?
Because people and process are where most of the trouble starts. BCG’s 2024 survey of 1,000 CxOs and senior executives found that around 70 per cent of the challenges companies face implementing AI stem from people- and process-related issues, 20 per cent from technology problems and only 10 per cent from the AI algorithms.
I’ve seen well-built automation go unused because nobody explained what it was for, and fairly ordinary setups produce results because the team understood them, trusted them and told someone when they broke.
People need to see the tool doing their own job. They need to know what it’s allowed to do and where their judgement still counts. They need somewhere to report that the output has started drifting, and they need to see something happen when they do.
The tell that this has gone wrong is silence. If nobody’s complaining about the AI, usually nobody’s using it. That’s what training is for, and it’s why we run it as part of the implementation.
What goes wrong most often?
In rough order of frequency:
- Buying before deciding. The subscription arrives before anyone has agreed what it’s for or who owns it.
- Treating it as a project. Projects finish. A change to how you operate has to be maintained.
- Ignoring the input. Bad data in, convincing rubbish out, at speed.
- Chasing sophistication. The most advanced option is rarely the one that pays back first.
- Relying on prompt skill alone. Individually skilled users produce individually good results that never become a process.
- The pilot that never ends. No end date, no decision, no result, budget quietly gone.
- Assuming headcount falls automatically. It usually doesn’t, at least not quickly, and planning as though it will creates problems that are harder to fix than the ones you started with.
Almost every one of these is a decision that didn’t get made.
Where should I start?
Pick one workflow that’s frequent, structured and slow. Work out what it costs you now in time or delay, so you have a number to compare against.
Decide who owns it before you buy anything, and agree what you’ll check and how often. Set a date by which you expect to see a change, and say out loud what that change will look like.
Then run it, look at the number and either keep it or stop. The businesses that get good at this are willing to switch something off when it isn’t earning its place, which only works if they decided up front what earning its place meant.
Do that two or three times and each one gets easier, because the discipline you built for the first is already there for the next. That compounding is the return. The time savings are the part you notice first.
If you’d like a second view on which job to start with, tell us about your workflows on a discovery call and I’ll come back afterwards with the one I’d pick first. If AI is already running somewhere in the business, the free AI risk assessment takes one of those uses through fourteen questions and ends on a decision.
FAQ
Where should a business start with AI automation?
Start with one frequent, well-structured, slow workflow, and measure what it costs you today. Quoting, enquiry handling and reporting are the most common first candidates because they’re repetitive and already have a shape.
How do I know if a process is ready to automate?
If you can describe the inputs, the steps and what a good output looks like, it’s ready. If the process only works because a particular person remembers how, sort that out first. Automating an undefined process just produces inconsistent results faster.
Do I need to fix my data before I start?
For that workflow, yes. You don’t need a data project across the whole business. The information feeding the process you’re automating has to be accurate and accessible, or the output will be confidently wrong.
Will AI automation reduce my headcount?
Usually not, and not quickly. The realistic outcome is that the same people handle more volume, or spend their time on work that needs them. Planning on the assumption that roles disappear tends to create problems that cost more than the savings.
How long before I see a return?
If the workflow was chosen well, within weeks, because the first candidates are deliberately small and measurable. If you can’t see anything within the window you agreed beforehand, that’s useful information too, and it’s a reason to stop.
General information only, not legal advice. Current as at October 2026: this area is moving quickly and the position may have changed by the time you read it.
