The dirty secret of enterprise AI rollouts: training events don't create habits. Here's what does.
Every AI rollout follows the same arc.
Month one: enthusiasm. People attend the training, explore the tool, try a few prompts. Engagement is high. Leadership is optimistic.
Month two: drift. Usage drops. The AI tool is open in a browser tab, but nobody's really using it. The habits never formed.
Month three: the tool is a line item on a vendor review.
Sound familiar?
Why This Keeps Happening
The problem isn't the tool. And it's not the team.
It's the training model itself.
Traditional AI training is event-based: a workshop, a webinar, a set of tutorials. It teaches people how to use AI in the abstract — but it doesn't teach them how to use AI in their actual workflow, for their actual tasks.
Research consistently shows that knowledge without application fades within days. The 70-20-10 learning model has been validating this for decades: 70% of learning happens through on-the-job experience, 20% through peer interaction, and only 10% through formal training.
One-time AI workshops are the 10% bucket. They're not useless, but they can't carry the weight we're putting on them.
What Actually Creates AI Habits
Habit formation requires three things:
- Context — the trigger has to be tied to an existing behavior
- Immediacy — the reinforcement has to happen at the moment of need
- Repetition — the behavior has to be practiced consistently over time
A training event delivers none of these reliably. It happens outside the work context, in a compressed time window, and then ends.
What does work: embedding AI assistance into the tools people already use, so that the help is available exactly when and where it's needed.
The Slack Advantage
For most knowledge workers, Slack (or Teams) is the primary place where work happens. Decisions get made there. Questions get answered there. Context gets shared there.
An AI coach that lives in Slack doesn't require behavior change. It meets people where they already are.
When someone is drafting a proposal in a Slack thread and can immediately ask Freddy "how should I frame this for a skeptical CFO?" — that's 70% learning. It's contextual, immediate, and repeatable.
Over time, these micro-interactions build real fluency. Not "I completed the AI training module" fluency. Actual, daily-use fluency.
What This Means for Your Rollout
If you're planning or evaluating an AI adoption program, ask one question: where does your team spend their day?
If the answer is Slack or Teams, that's where the AI coaching should live — not in a separate portal, not in a dedicated app, not in a quarterly training calendar.
The tools that create lasting AI adoption are the ones that fit into existing workflows, not the ones that demand new ones.
BrainGem's Freddy is an AI coaching layer for Slack. If you're working on AI adoption and want to see how this works in practice, visit braingem.ai.
![Defluffer - reduce token usage 📉 by 45% using this one simple trick! [Earthday challenge]](https://media2.dev.to/dynamic/image/width=1000,height=420,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fiekbgepcutl4jse0sfs0.png)
