When They Ask for AI Training, Hand Them This Intake First

When They Ask for AI Training, Hand Them This Intake First

The request comes in a form every L&D leader recognizes. It sounds like this: “Can you build us a course on Copilot?” Or: “Our sales team needs AI training by Q1.” Or the bluntest version: “Just make it quick.”

Last year, most of those requests would have been about a new system or a compliance update. This year, they are about AI, and they are arriving faster than anyone can build for. The pressure to say yes quickly is real.

Say yes too fast and you end up doing the most expensive kind of work there is: building training for a workflow nobody has redesigned.

The data is getting blunt about this. Deloitte’s 2026 research found that 84% of organizations using AI have not rewired a single role or workflow around it. They bolted the tool onto the old job and called it adoption. McKinsey’s 2026 numbers complete the picture: almost 90% of companies now use AI somewhere, but only about 6% see material impact on earnings.

If the work has not changed, training people to use the tool inside the old work does not create value. It creates activity. That is why the next training request on your desk deserves five questions before a single course is designed.

The intake: five questions before you build anything

These are the questions a stakeholder has to answer, with you in the room, before the request becomes a project. Write their answers down. The answers that come back will tell you whether this is a training problem, a work redesign problem, or a request that should not have been made at all.

1. What performance outcome is this supposed to change?

Not the completion target. The business measure. “Forty salespeople will complete the course” is not an answer. “Proposals reach clients two days faster” is an answer. If the stakeholder cannot name the outcome in one sentence, you do not have a training request. You have a conversation that has not happened yet. Schedule it and do not start building.

2. What has changed, or must change, about the work itself?

AI rarely fits the old workflow. It usually replaces parts of it. Before training anyone, the team needs to know which parts of the job stay human, which parts AI assists with, and which parts should no longer be done the way they were done. If nobody has worked that out, your course will teach people to use a tool inside a process that should have been rebuilt. Training goes second. Redesign goes first.

3. Who decides whether the AI’s output is good?

This is the question almost nobody asks, and it is the one that sinks most deployments. When AI drafts the proposal, the customer email, or the learning module, someone with standing has to judge whether the output meets the standard. Name the person, or name the review step. If the answer is “the tool will handle it,” the request is not ready. No reviewer, no launch.

4. What will stop being done to make room?

Every AI request adds work before it saves work. Someone has to learn the tool, change habits, and check output that used to flow through automatically. If the team is already at capacity, the honest question is what gets cut. Requests with no cut list are requests where the savings exist only on paper. Get the cut list in writing before you commit to a timeline.

5. How will we know it worked, and when?

Pick the measure from question one, set a date, and name the baseline. “We will check back in March” is not a plan. “We compare proposal turnaround in Q1 against the Q4 baseline, target two days faster” is a plan. An evaluation date agreed in advance is also your protection. It is much harder for a stakeholder to call the training a failure in April when they agreed in November on what success would look like.

The decision rule

Five questions, and here is what the answers buy you.

If the stakeholder cannot answer the first question, treat the meeting as consulting, not as intake. Your job is to help them find the outcome before anyone talks about training. That is diagnosis, and it is the highest-value work L&D does.

If they can answer the first question but not the second or third, the request is premature. Send it back with the unanswered questions attached. That is not obstruction. It is the only way to keep the team from building the wrong thing on schedule.

If all five have answers, you have a real project. Build it.

Notice the sequence. Extraction of what the work requires comes before design, and design comes before generation. That is the EDGE sequence from AI for Learning and Development: Extract, Design, Generate, Evaluate. Most teams start with generation, which is why they end up with faster content and unchanged performance. The intake form forces the correct order on everyone, including the stakeholder who wanted it quick.

Going further

The method behind this series is the EDGE framework, Extract, Design, Generate, Evaluate. My book AI for Learning and Development works the full system with the prompts and workflow templates:

https://www.amazon.ca/dp/B0D5YJ1QQG

The reframe worth keeping

A training request is not a diagnosis. It is a symptom. Someone saw a problem, reached for the tool they know, and named training. Half the time they are right and half the time they are not, and you cannot tell which half you are in until you ask.

L&D has spent years absorbing the cost of skipped diagnosis. Every course built for the wrong problem was a bet made without checking the odds first. AI raised the stakes on that habit, because the requests are more frequent, the tools are more seductive, and the redesign questions underneath them are harder than anyone wants to admit.

The intake is not paperwork. It is where the series line becomes operational: your people do not need more AI training. They need to know how their work changes. The first place they learn that is in the five questions you ask before you say yes.

Print the intake. Use it at the next request meeting. Watch how fast the room sorts real projects from borrowed ones.

Dr. Ravinder Tulsiani is the author of AI for Learning and Development: Faster Content Isn’t Better Learning and Solve the Right Problem: A Practical Guide for L&D. Take the AI-Ready L&D Function Assessment.

dummy-img

What Do You Train People For When Their Jobs Keep Changing?

Dr. Ravinder Tulsiani Nov 18, 2026 7 min read

How workforce leaders can build meaningful capability plans when AI…

dummy-img

Why Organizations That Invest Heavily in Learning Can Still Struggle to Adapt

Dr. Ravinder Tulsiani Nov 17, 2026 8 min read

By Dr. Ravinder Tulsiani A sales team finishes training on…

dummy-img

AI Training Won’t Fix a Workforce Strategy That Hasn’t Changed

Dr. Ravinder Tulsiani Nov 16, 2026 10 min read

By Dr. Ravinder Tulsiani The budget review six months in…

Get practical insights on workforce capability, AI readiness, and learning strategy — delivered on LinkedIn.