Training for every tool does not scale
When a new GenAI tool arrives, the instinct is to train for it. Evaluate, decide, roll out, train. The vendor provides materials. Attorneys learn what the tool does, how to access it, and what risks to watch for.
But when every tool gets its own training path, the burden shifts to the attorney to assemble the bigger picture — which tool to use when, which guidance applies where, and whether what they learned in a Copilot session also applies in a legal research tool.
That is where GenAI overload begins to show up. Not necessarily as resistance, but as fragmentation.
That is a lot to ask. And it compounds quickly: more tools require more training, more training requires more updates, more updates create more communication.
More becomes busy, not clearer.
The missing layer: training for use
Most GenAI tools share the same underlying behaviours. They respond to instructions. They work better with context. They can summarise, draft, classify, and structure thinking. They require review. They can sound confident when wrong.
Training for use means training in those behaviours — not in a specific tool, and not in a specific workflow. It is about building a mental model of how GenAI works, so that the learning travels. What does good input look like? What are the failure modes? How do I know when to trust the output and when to push back?
Once attorneys have that foundation, learning becomes cumulative. Workflow training — how to run a contract review, how to draft a client alert, how to build a chronology — can sit on top of it and actually land. Without it, workflow training becomes mechanical: attorneys learn which steps to follow without understanding how to adapt when the output is weak or the tool behaves differently than expected.
This is the distinction worth holding clearly. Training for workflows is valuable. But it is not the same as training for use — and it cannot substitute for it. A firm can have excellent workflow guidance and still leave attorneys without the foundation they need to exercise judgment.
The questions that matter at the foundation level are not platform-specific and not workflow-specific:
- What is this type of tool good at?
- What input improves the output?
- What should I never outsource to it?
- How do I review the result?
- How to combine tools for the best results?
Tool-specific training still has a place, as does workflow training. But that specificity should sit on top of a shared foundation — not replace it.
Design an onboarding sequence, not a pile
The practical question is simple: where should an attorney who isn’t using GenAI today begin?
That question is harder to answer than it looks — and in many firms, the answer has emerged gradually rather than intentionally. The result is that curious attorneys have to reverse-engineer a learning path from a collection of trainings, recordings, tool announcements, and office hours invitations.
Before designing anything new, map what already exists. List every GenAI-related training session, guidance document, pilot, demo, and communication currently available. Then add one column: the time required from the attorney.
That exercise is often revealing. The same basic concepts are repeated across sessions. Tools were introduced in isolation with no guidance on when to use them over alternatives. A support system that grew by addition — a new tool, so a new training; a new risk, so new guidance — until it became a patchwork that is hard to navigate and hard to maintain.
This is often how GenAI overload develops: not through one bad decision, but through many reasonable additions that never become a clear path.
From more activity to clearer progression
The next phase of GenAI adoption won’t be won by offering more. At some point the question changes — not what else should we provide? But how does this fit together for the person trying to use it?
Attorneys are not resisting GenAI because they want another demonstration. Many are trying to understand where it belongs, how to use it responsibly, and which option deserves their limited attention.
The firms that make progress will be the ones that reduce the sense of fragmentation — by training for use before training for tools, by making the basics transferable, and by designing onboarding as a sequence.
Adoption isn’t created by activity. It’s created when the next step is clear enough to take.