Before you automate a task, ask whether you need it.
Artificial intelligence can help us work faster. First, it’s worth looking at what we do, why we do it and where people’s time is best spent.
Imagine your team spends two hours a week preparing a report. You introduce an AI tool and get a first draft in ten minutes.
It sounds like a success.
Three questions are still worth asking:
- Does anyone use this report to make a decision?
- How much time and effort does it take to review?
- Did anyone need it every week?
The example is hypothetical, but the decision is real: before speeding up a task, find out what purpose it serves.
Perhaps you need a faster report.
Perhaps three indicators would be enough.
Or perhaps you could stop producing it altogether.
1. Start with the outcome, then consider the tool
When a task has been part of the routine for years, it’s easy to mistake habit for necessity.
To review it, bring together the people who do the work and those who use the result. Ask:
- What decision does it inform, what need does it meet or what problem does it solve?
- Who uses what we produce, and how?
- What would happen if we stopped doing it for a month?
- Is there an obligation or a control we need to keep?
- Could we achieve the same result in fewer steps?
That last question often opens up more interesting possibilities than starting with a search for an app.
Take the report. If the recipient only needs to know which orders are late, a full document every Friday may be unnecessary. An alert when something goes wrong might be enough.
Check with the people involved. Even a tedious task, or a report with few readers, may help prevent errors, keep an audit trail or meet an obligation.
2. Decide what to stop, simplify or automate
Once the purpose is clear, it’s easier to choose the next step.
Stop doing what no longer adds value
Consider dropping a task when it no longer serves its purpose, duplicates another activity or produces something nobody needs. Before retiring it, check that someone still covers any underlying responsibility.
For example, stop producing a second version of a report if both recipients can use the first.
Simplify what still serves a purpose
Sometimes there are unnecessary steps, fields, formats or approvals.
You could cut a report down to the information that informs a decision, avoid entering the same data twice or check whether every approval is necessary.
Simplifying first avoids carrying the same complexity into a new tool.
Automate once the task is clear enough
If a task has clear rules, reliable inputs and a result you can verify, it may be a good candidate for automation.
Some automation only needs rules. Rules can send a reminder when a deadline arrives or transfer data between systems. AI can help summarise text, prepare drafts or explore information, but its answers can contain errors and need a level of review that matches the risk.
Let the work guide the choice. A more sophisticated tool won’t necessarily be more useful.
3. Decide where people need to stay involved
In some tasks, the human contribution lies in interpreting the context, talking things through, weighing alternatives or taking responsibility for a decision.
AI can help prepare a development conversation. Listening to someone, understanding their expectations and agreeing the next steps takes human involvement.
AI can also draft a report. Someone needs to check that it reflects what happened and that the conclusions are justified.
Be clear about who reviews the work, who decides and when an extra check is needed. Otherwise the time saved on production can reappear later as corrections, questions or duplicated effort.
4. Look at the whole role as well as the individual task
Changing one activity can affect several people’s work.
Someone who stops gathering data may need to spend more time interpreting it. Someone receiving automatically generated drafts will need criteria for assessing them. And if an administrative burden is reduced, someone has to decide how to use the time it frees up.
Review the work with the people doing it and with their managers:
- What outcomes is the role expected to deliver?
- Which responsibilities change and which stay the same?
- What skills will the new way of working require?
- What support or training will people need?
Job analysis helps connect decisions about technology to people’s actual responsibilities. It means people can see how the new tool fits into their working day.
5. Start small and measure the whole process
- Choose one clearly defined task where you can reverse the change.
- Decide what you want to improve and what must not get worse before you start.
- Record how the work is done today, then compare several similar cases using the new approach.
- Include some routine cases and at least one exception.
Look at four things:
- Total time: preparation, execution, review and corrections for everyone involved.
- Quality: errors, omissions and work that has to be repeated.
- Usefulness: whether the result actually helps the person receiving it.
- Privacy and control: which data is used, where it is processed and who can access it.
For an initial trial, use fictional data or information you’re authorised to use, and tools approved by your organisation. Agree who will oversee the result and how to return to the previous process if something goes wrong.
In our example, the ten minutes it takes to create a draft tell only part of the story.
Look for improvement across the whole process.
A question for your next meeting
Ask your team: “Which task would you review before automating it?”.
Listen to the answers. Choose one task. Clarify its purpose and try a small change before rolling it out.
AI offers opportunities to improve work. Making the most of them means deciding what’s worth keeping, what we could do differently and which skills we need to develop.
How is work changing for your teams?
At Esferical, Job Analysis and Talent Optimisation offer two ways into that conversation: understanding the work and identifying where people can contribute. Let’s talk about the responsibilities and skills your organisation needs.