I was reminded of this recently while discussing outbound workflows.
The idea sounded reasonable on paper, but it relied on letting the executive do a bit of research before sending.
In reality, that’s exactly where things break.
The moment a workflow requires research, momentum disappears.
Not because the output is bad, but because the cognitive load is too high.
Thanks for reading,
Robbie Allen
Founder & Managing Director
Automated Consulting Group
PS: If AI adoption is stalling inside your org, hit reply. The issue is usually workflow design, not capability.
Key Takeaways:
AI adoption fails when workflows add work instead of removing it.
Dashboards create insight, not action. “Review + send” beats “Analyze + decide” every time.
Cognitive load is the real bottleneck to AI adoption, and the best AI workflows feel like progress, not software.
AI Adoption Is a UX Problem, Not a Technology Problem
Most teams talk about AI adoption as a change management issue, but it's actually an issue of design and user experience.
If an AI workflow asks someone to:
Log in to a tool
Review a dashboard
Interpret results
Decide what to do next
…it’s already lost.
Compare that to an inbox draft. There's no login, interpretation, or prioritization. Just one question: send or delete?
Executives don’t lack interest. They lack spare attention.
The highest-adoption AI workflows I’ve seen don’t require learning anything new. They show up where work already happens and reduce the number of decisions required to act.
Dashboards create insight, but inboxes create action.
A dashboard might tell you what to do, but it still leaves the hardest part to the human: deciding when and whether to act.
An inbox draft collapses that distance.
Designing for Momentum (What Actually Works)
If you’re building or deploying AI internally, here’s the test I use:
Can someone act in under 30 seconds without thinking?
If not, adoption will stall.
And in designing the workflows, these are the patterns we see work well for CEOs, sales leaders, and ops teams:
1. Collapse the workflow to one decision
Every extra decision is a tax on adoption. The highest-usage AI workflows reduce action to a binary choice:
Send or don’t send (for example, a drafted reply in the inbox)
Approve or delete (for example, a weekly status update draft)
Accept or ignore (for example, AI detects customer churn signals and proposes outreach)
Once a workflow requires prioritization, comparison, or judgment calls, it competes with higher-stakes work and loses.
Design rule: If the human has to decide what to do, adoption is already at stake. The AI’s job is to propose the action.
2. Ship output, not insight
Most AI tools are built to surface information like performance metrics, recommendations, and opportunity lists, but leaders don’t struggle with knowing what’s possible.
They struggle with turning intent into action.
The AI workflows that stick don’t say: “Here are 12 opportunities ranked by likelihood.”
They say: “Here’s the next thing to do.”
Design rule: If the output doesn’t look like a finished artifact, adoption will stall.
3. Put AI where work already happens
Every new surface area is friction. If your AI requires a new login, a tab switch, or a new habit, usage will decay over time.
Email, calendar, Slack, and CRM systems already have built-in attention. AI that shows up inside those systems borrows that momentum and becomes a more natural part of the work.
Design rule: Never ask users to “go check” an AI tool. Make the AI come to them.
4. Treat “good enough” as the target
Executives do not need perfect AI output, just something that beats not making a decision. A draft that’s 80% right and ready now will outperform a perfect one that requires more thinking later.
This is especially true for outbound email drafts, follow-ups, internal updates, and routine decisions.
Design rule: Design for speed to action, not theoretical quality.
The Takeaway for Leaders
When we work with leadership teams, we don’t start by asking: “What could AI do here?”
We start by asking:
“Where does work stall today?”
Then we design AI to remove friction at that exact point.
That’s how AI becomes operational instead of experimental, and how adoption becomes a given.
– Robbie
P.S. I’m curious. Hit “Reply” and tell me where your AI adoption is currently lower than you’d like it to be.