How to make time for AI learning when the team is busy

To make time for AI learning in a busy team, agree which existing work will change before you put practice on the calendar. Choose a useful task, estimate the extra effort involved, and name the person who can approve the tradeoff. Keep delivery expectations consistent with that decision.

You can start with a small attempt. Two colleagues might try an approved tool on a recurring task, with someone experienced checking the result. The allocation needs to cover their preparation and review as well as the time spent using AI. It also needs a review date, so a promising experiment can continue and an unhelpful one can stop.

This article gives you a capacity example and an agreement to adapt. The broader guide to helping people adopt AI at work covers access, learning, trust and support.

Choose the work people will learn through

Begin with a task the team recognizes and a question worth answering. Suppose a retailer's warehouse coordinators want an approved AI tool to draft the weekly delivery-exception summary: the orders that arrived late, with missing items or damaged goods, and what each store needs to know. They would use approved warehouse records, with a supervisor checking that affected orders and unresolved problems remain visible.

Define an acceptable result before allocating time. For this summary, every late, short or damaged order must trace back to the warehouse record. The supervisor checks that quantities and confirmed arrival dates are accurate, and that unresolved orders have a next action. If you have not chosen the task yet, use the task-discovery worksheet.

In her CIPD analysis of line managers and skills development, Laura Overton emphasizes the manager's role in giving people opportunities to apply learning after training. She recommends agreeing on outcomes that matter to the business. For your plan, that means explaining what the team hopes to learn about its work, beyond completing a course or trying a feature.

Estimate the additional effort for the first attempt:

• Preparation. Confirm access, find suitable material and understand the relevant guidance.
• Practice. Try the approach, inspect the output and make corrections.
• Review and reflection. Get appropriate feedback and save what another colleague would need to repeat the attempt.

Separate that extra effort from the task's normal delivery time. If someone already has an hour to produce the summary, do not count that same hour again as new learning capacity. Equally, do not assume that an unfamiliar method will fit inside it. Keep a way to finish the task using the existing approach if the experiment does not work.

Make the capacity tradeoff explicit

In this warehouse example, the manager wants two coordinators to test the delivery-exception summary. Each needs 90 minutes of additional preparation, practice and reflection. An experienced reviewer needs 30 extra minutes to examine the attempts with them. The total is 210 person-minutes, or 3.5 person-hours.

The manager finds two changes for this week's plan. First, the two coordinators and warehouse supervisor can skip a 30-minute delivery-status meeting because the order updates are already available in their normal tracker. No replacement report is needed. Second, the retail operations manager who receives the summary agrees to drop its optional presentation charts this week. Preparing those charts would have taken each learner an hour; the required figures and normal accuracy checks remain.

Canceling the meeting releases 90 person-minutes. Dropping the charts releases another 120. Together, those changes cover the 210 minutes required, with the time available to the people doing the work.

Check the assumptions before changing the schedule:

• Can the meeting be removed without losing a necessary decision?
• Does the report's recipient agree to the simpler presentation?
• Can the reviewer join when the learners are ready?

If the replacement arrangements create extra work, add that effort before saying the time is covered.

You may have no meeting or optional report to remove. Other choices include moving a delivery date with its owner, reducing the scope of the attempt, or arranging additional cover. A sponsor must approve changes beyond your authority. Put the competing commitments in front of that person rather than leaving employees to reconcile them after hours.

Write down the manager's commitments

Keep the agreement in the team's existing planning system. It should be short enough to use when another request arrives. Record:

• The learning question. Name the task, the people involved and what would make the attempt useful.
• The allocation. Include each person's extra time and the point when a reviewer is needed.
• The changed work. Say what is removed, reduced, moved or covered, and who agreed to that change.
• The conditions for use. Confirm the approved tool, suitable information and the standard the output must meet.
• The review. Set a date to examine the result, actual effort and delivery consequences.
• The interruption rule. Name who can displace the session and how a new time or revised scope will be agreed.

Make the same change in the delivery plan. A learning agreement and an unchanged workload forecast give employees conflicting instructions. Check that the report recipient and the person allocating the team's delivery work know what has changed.

A manager's expectations also affect what people feel able to do during practice. In a public Reddit discussion, one Reddit user described allowing AI use while continuing to expect good code quality and clear communication. The commenter, who described managing a team, also said they did not expect everyone to produce ten times as much. Their employer and team size were not stated. The account illustrates a choice about expectations, rather than evidence that one approach works everywhere.

Be equally clear with your team. An experiment can reveal that a task needs more checking or that the existing method is better. Ask for that finding honestly; do not make promised time savings the price of receiving learning time.

Work with the delivery constraints you actually have

A desk-based team with flexible internal deadlines has different options from a staffed service queue. Before choosing a session, work through the constraint that applies:

• Customer coverage. Identify who will handle incoming work, whether they have room, and what happens if demand rises. Staggered sessions may be more workable than taking everyone out together.
• Billable commitments. Ask the person who owns the account and workload target how the learning allocation will be treated. A manager's invitation does not by itself change the expected billable work.
• Shifts and part-time schedules. Put practice and access to help within the participants' working arrangements. A session available only to one group leaves others without the same opportunity.
• An immovable deadline. Reduce or reschedule the attempt with a named owner and date. Repeated postponement calls for a new capacity decision. When several programmes need the same people, map their overlapping demands before choosing a new date.

These are planning questions, not a promise that every constraint has an easy workaround. Where no delivery commitment can change and no cover is available, take the unresolved allocation to the sponsor. A smaller supported attempt may be sensible; quietly adding the full programme to everyone's existing work is not an agreed plan.

People supporting colleagues need their own allocation too. If a champion will prepare examples, answer questions or follow up between sessions, use the champion workload diary to include those duties.

Review the result and the workload together

At the agreed review, bring the finished attempt and the original capacity agreement. Use the team's regular AI review conversation if it already provides a suitable place.

Work through the decision in order:

1. Check what happened. Did the planned work actually move, and did people get the time? Record extra preparation, support or rework that the estimate missed.
2. Examine the output. Can the learner explain what they checked and what they would change? Decide whether the result meets the task's standard.
3. Choose the next commitment. Continue, narrow, change or stop the approach. Agree the capacity for any next attempt rather than assuming it is now free.

If delivery slipped, find out which assumption failed. The learning estimate may have been too small, the removed work may have returned, or an unrelated demand may have arrived. If the output was useful but the workload was unsustainable, both findings belong in the decision.

For your next planning conversation, bring one task, the extra effort it needs and a specific proposal for what changes. Leave with a named approver and a review date. That makes learning time a commitment the team can act on.

Once a task becomes faster, check whether AI has reduced workload or raised expectations before assigning the recovered capacity to new targets.

Questions about making time for AI learning

How many hours a week should employees spend learning AI?

Budget AI learning around a specific task, including preparation, practice and feedback. The evidence discussed here does not establish a weekly allocation that fits every team. Two warehouse coordinators learning to draft a delivery-exception summary may need different support from experienced colleagues improving an approach they already use.

Start with one bounded attempt and estimate the extra time each participant needs. Include the reviewer, agree which existing work changes, and compare the estimate with actual effort afterward. The person-by-person capacity example shows how to check that released time reaches the people who need it.

What if nothing in the current workload can move?

A manager needs someone with authority to resolve the conflict between AI learning and existing delivery commitments. Take that person a specific proposal: who will practise, what they will try, how much additional effort it needs, and which work could move or shrink. An invitation to learn does not create capacity while every existing commitment remains fixed.

If no tradeoff or cover can be approved, narrow or reschedule the attempt with a named owner and review date. Do not leave employees to make up the difference through unplanned extra work. Use the questions for customer coverage, billable work and shifts to explain the constraint that needs resolving.

Can learning time coexist with billable targets?

It can, if the person responsible for workload and client commitments agrees how the learning allocation will be treated. Billable targets are the amount of client work employees are expected to charge for. Setting aside practice time without adjusting those expectations can leave someone with the same client workload and less time to complete it.

For example, a design agency could move an internal portfolio review so designers can practise drafting descriptions for product photos, including time for a colleague to check each description against its photo. The account manager would confirm that the client delivery date still holds. That only works if the people who own both commitments agree.

Record the changed work and approver in the manager's learning agreement; assess possible future efficiency after the attempt rather than counting it as time already available.

What should a manager do when a learning session is canceled?

Find out what displaced the AI practice session, then agree a replacement time or a smaller attempt with the people involved. Recheck the reviewer's availability and the work that was supposed to move. Restoring a calendar invitation alone may leave the original workload conflict unresolved.

When the same delivery pressure repeatedly cancels practice, return to the approver with that pattern and ask for a new capacity decision. The team may need cover, a narrower learning question or a different date. The review of output and workload explains what to examine before committing to the next attempt.