A task that keeps moving to tomorrow is worth examining. It may be tedious, poorly defined, under-resourced or no longer necessary.
Find out why it keeps slipping before deciding to automate it. A tool can help with part of the work, but it cannot make an unnecessary task worth doing.
For a first project, choose something narrow enough to understand and useful enough to justify maintaining.
Make a short list from actual work
Ask the people doing the work which recurring tasks are difficult to finish. Include the reason they give, how often the task happens and who needs the result.
Look at a recent example rather than relying only on a general impression. Where did the time go? Was someone waiting for a decision? Did the task produce something another person used?
A report nobody reads and a customer enquiry nobody owns need different responses. The report may need to stop. The enquiry needs a dependable handoff.
Compare the candidates
For each task, ask:
- Usefulness: who needs the result, and what do they do with it?
- Effort: how much work does a normal case require, including corrections?
- Clarity: can we describe a correct result and recognise a failure?
- Information: are the necessary inputs reliable and appropriate to use?
- Risk: what happens if the task is wrong, late or duplicated?
- Ownership: who will maintain the process and handle exceptions?
You do not need a complicated scoring formula. Use the answers to compare the practical trade-offs and identify what is still unknown.
Check the simpler fixes
A missing owner may need an assignment rule. A confusing form may need a better question. A recurring reminder may be possible with software the business already uses.
AI may help when a bounded part of the task involves interpreting varied information or preparing a draft. Include the human review in the plan.
For a hypothetical electrical contractor, the office may spend time rebuilding the same weekly enquiry summary. The first step could be standardising the fields and confirming who uses the report. Only then is it useful to test a tool that prepares the summary.
A fast summary of inconsistent records can still leave the manager with the same uncertainty.
Run a trial with a clear stopping point
Choose representative examples and agree the time and spending limits. Define which information may be used and which actions need approval.
Include an incomplete or duplicate example. Check what happens when the tool cannot produce a dependable result, and keep a manual way to finish the work.
Record the review and maintenance effort alongside any time saved. A process that moves work from the office to an unseen checking queue has not necessarily reduced the total burden.
Decide what the trial taught you
At the review, compare the result with the original reason for choosing the task. Did the intended person receive useful work? Were errors or delays reduced? What remains unresolved?
You may keep the approach, change the scope or choose a simpler fix. Record the decision and the person responsible for the next step.
If you can name a task that keeps being postponed, tell us what it is and who needs the result. We can help assess whether it is worth simplifying or automating.
