I recently saw an argument that enterprises should start with AI training, get employees using AI, then bring in an FDE.
It sounds reasonable, but deserves closer examination.
First, AI training is not a standard step in the FDE workflow. From the approaches I have seen, the usual path is understanding client goals, interviewing key roles, reconstructing workflows, identifying pain points, assessing AI opportunities, narrowing an MVP and delivering it.
There is no fixed step called “train everyone first.”
In practice, companies sometimes choose training before hiring an FDE. Leaders want employees to understand AI and learn the tools, then identify processes worth changing.
That is reasonable. But training is not implementation. What helps an FDE is whether training leaves behind real business scenarios.
Why generic AI training rarely delivers implementation
Many sessions become engaging but ineffective introductions. A trainer explains prompts, demonstrates general tools and shows impressive examples. Employees leave inspired but do not know how to use them in their own work.
The problem is not a lack of enthusiasm. Training has not entered the business workflow.
What employees often need is not knowledge of what AI can do, but answers to these questions:
- Which step in my work could change with AI?
- What are its inputs and outputs?
- Are there historical examples?
- Who reviews the generated result?
- Which system, spreadsheet, document or collaboration flow should it return to?
- What is the risk of getting it wrong?
Without these answers, the outcome can be:
Learn a few AI tools -> try them briefly -> return to old habits -> no real organizational adoption
“Training first, FDE second” is not a reliable methodology. A better description is:
AI training can improve organizational understanding, but cannot replace business diagnosis and implementation by an FDE.
Three kinds of enterprise AI training
I divide training into three types.
Type one: generic tool training
This is the most common. It usually covers:
- AI tools
- Prompt writing
- General office productivity examples
- Writing, summaries, translation, presentations and image generation
Its value is lowering the knowledge barrier: AI is neither mysterious nor reserved for technical teams. But it provides limited help to an FDE, who must still ask:
- Which workflow hurts most?
- Who initiates, processes and reviews it?
- Where is the data?
- Are there real examples?
- Are inputs and outputs clear?
- Can it become an MVP?
This is an introduction to AI, rather than preparation for implementation.
Type two: collecting scenarios after training
This starts to become useful. Departments are asked to submit scenario lists rather than simply learn tools:
What are our three most repetitive activities?
Which three workflows take the most time?
Where do errors or rework occur?
Do we have spreadsheets, documents, chats, tickets or screenshots?
Who could review an AI-generated suggestion?
Training is not yet part of the FDE workflow, but it produces input materials. The FDE can evaluate:
- Which problems are real rather than imagined?
- Which occur frequently?
- Which have data or examples?
- Which support human review, and which carry excessive risk?
- Which can become an MVP in one or two weeks?
This helps because it produces business material that can be diagnosed.
Type three: training as a scenario workshop
This is the strongest form. It looks less like traditional training and more like an AI scenario workshop.
Participants bring actual work: a recent order, report, batch of customer conversations, ticket flow, quotation or delivery document. Together they examine:
- How does the workflow run?
- Which step is most frequent?
- Which takes the most time?
- Which creates errors?
- What are the inputs and outputs?
- Are there historical examples?
- Should AI suggest, classify, draft, retrieve knowledge or merely assist checks?
With an FDE or someone experienced in business implementation involved, the workshop becomes early business diagnosis. Training and delivery begin to overlap, but the principle is:
The FDE turns training into a mechanism for discovering scenarios.
What the FDE actually receives
It is not simply employees who have studied AI. Useful inputs are:
- Actual business workflows
- Repetitive frontline work
- Recent concrete examples
- Spreadsheets, documents, screenshots, chats and tickets
- Clear inputs and outputs
- Results that people can review
- Pilots a leader or manager is willing to support
Without these, the FDE still needs fresh interviews and clarification. Training's value is not how lively the session was or whether people can write prompts, but whether it leaves usable outputs:
Department scenario lists
Real business examples
Workflow or system screenshots
Prioritized recurring pain points
Reviewable AI candidate scenarios
Business owners or pilot departments
These are what the FDE can carry forward.
A more accurate relationship
I now see it this way:
AI training = organizational understanding
Scenario collection = inputs for an FDE
FDE = business scenario diagnosis and implementation
If a company already has a clear problem, leadership authorization, frontline examples and system data, an FDE can start directly without general training.
If it only says “we want AI productivity” and has no understanding of applications, introductory training is reasonable. But its endpoint should be collecting real business scenarios, not merely teaching tools.
Otherwise, it produces employees who know AI is impressive but do not know how to use it in their jobs.
My conclusion
Training has value, but is not a necessary prerequisite for FDE work. What matters is turning vague AI interest into business material that can be diagnosed.
Tool-only training mainly addresses understanding. Training that captures scenarios supplies inputs for delivery. A scenario workshop can become part of early FDE diagnosis itself.
Training shows employees that AI may be useful; scenarios let the FDE judge where investment is worthwhile; an MVP demonstrates whether AI actually helps this enterprise.
Thanks for reading.