The licences are active, but the team still uses its old workflow. AI adoption needs a useful task, suitable inputs, clear review responsibilities and enough time to learn. A training session cannot compensate for a tool that adds checking and administration without a worthwhile result.
Photo by Microsoft Edge on Unsplash.
Begin with the user's actual difficulty
Ask staff to demonstrate a recurring task and explain what makes it slow or frustrating. Choose a bounded use they recognise as useful. Invite the people who will review outputs into the design, because their effort is part of the new workflow.
Teach the whole task
Practise choosing approved inputs, generating a draft, checking it against the source, correcting it and routing exceptions. Include an incomplete or misleading output. A successful exercise should show when the tool should not be used, not only how quickly it can answer.
Make feedback easy to act on
Record the task, result and problem encountered.
Name the person who can change the workflow or instructions.
Separate training gaps from tool or process problems.
Explain what was changed after feedback.
Keep a supported alternative while the pilot is assessed.
An illustrative adoption problem
Account managers receive an AI summary but must search several systems to verify it. The team may reasonably avoid the tool. Bringing the source material and checking steps together could matter more than another session on prompt wording.
Related guidance: guide to mapping the work before changing it.
Track acceptable completed work, review effort, corrections and user experience. Login counts or generated drafts alone do not show that the change helps. Expand only when the evidence supports the operating model and users can maintain it. Hili Consulting can help connect AI training with a workflow that serves the people doing the work.
