Questions we answerAI and automation

What should I automate first?

The most frequent, most repeatable, best-understood, lowest-stakes thing you do. Not the most annoying one, and not the one with the best demo.

Start from this question

Free · a few minutes · starts from “I think AI could help

The short answer

Start with work that is frequent, genuinely repeatable, already written down, and low-stakes if it goes wrong. That combination is what makes a first automation succeed quickly and cheaply enough to earn the second one. Do not start with the most irritating process — irritation usually tracks complexity and judgement, which is exactly what automates badly. And check what the systems you already pay for can do before building anything: their automation is already governed, logged and supported.

The first automation matters more than its size, because it sets whether there is a second one. A first attempt on the hardest process in the business usually ends with everyone concluding that automation does not work here.

So the selection criteria are not about value. They are about the odds of finishing.

Is it really this?

Worth checking before you spend anything. These are the signals that tend to tell the two apart.

Signs it is

  • You can name a process that happens daily, the same way, and is written down.
  • Getting it wrong is embarrassing rather than expensive.
  • One person could describe it end to end in five minutes.

Signs it's something else

  • Nothing you do is genuinely repeatable, and the load is judgement-heavy.
  • The candidate processes all change every few weeks.
  • The pressure is volume of valuable work rather than administration around it.

What's usually behind it

  • A general sense that too much of the week is manual, without a specific target.
  • A tool that has been bought and now needs something to run.
  • One process that annoys everyone, which is therefore assumed to be the right place to start.

What people usually get wrong

  • Starting with the hardest process because it would save the most.
  • Choosing by how annoying something is rather than how automatable.
  • Ignoring the automation already included in the tools you pay for.

The options, and when each one fits

In no particular order. Which of these fits depends on your situation, not on the question you typed.

  • The frequent, simple, documented one

    Almost always, for a first automation.

  • Whatever your existing systems already do

    The work lives inside a system with automation features you have not used.

  • Nothing yet

    No process is written down. Write one down first.

When doing nothing is the right call

If nothing meets the bar, that is a finding rather than a failure. The right first step is then documenting one process properly, which is a week of work and makes the next answer obvious.

What the diagnostic asks

  • What the work is and whether it is mostly repeatable.
  • How often it comes round and how much judgement it needs.
  • What a mistake costs and how sensitive the information is.
  • Whether software already exists for it.

“I don't know” is always an answer. Nothing is assumed on your behalf.

The routes it weighs

Every diagnostic weighs all ten routes. These are the ones that tend to matter for this question — a description of the rules, not a promise about your answer.

  • Process improvement

    Change how the work flows — remove steps, name an owner, fix the hand-offs.

  • AI Assist

    Keep a person in charge and give them AI for drafting, summarising, checking or analysis.

  • AI Agent

    Delegate a defined, repeatable workflow to software inside boundaries you set, with human oversight.

  • Do nothing yet

    Record the issue, set a revisit date, and decide what would change your mind.

What changes the answer

  • Frequency is the strongest single factor — daily work pays back setup that weekly work does not.
  • Judgement along the way changes the shape of what you build rather than whether to.
  • An existing system that already does most of it is usually the actual answer.

Get your own read

A handful of short questions, then the strongest route for your situation, what's also credible, what we still don't know and what would change our view. No account, no setup.

Related questions