Stop Waiting for the AI Roadmap
If your org isn’t already making AI tools available to your teams with appropriate guardrails — what are you waiting for?
I keep hearing the same thing from engineering leaders and security managers: “We’re working on our AI strategy.” “We need to figure out the right way to roll this out.” “We’re waiting on legal/compliance/procurement to finalize the policy.” I’ve written about the adoption stages — and most of these orgs are stuck between Stage 1 and Stage 2, not because of the tech, but because of the bureaucracy.
Meanwhile, your talented engineers are either using AI tools anyway (without guardrails), or they’re falling behind because they’re following the rules and waiting for permission that never comes.
Both of those outcomes are worse than what I’m proposing.
Just Open the Budget
Here’s my argument, and I know it sounds uncomfortable to anyone who’s ever had to justify a line item: open up the AI budget with no clear target or implementation plan.
Yes, really.
This isn’t how we normally do things. Normally there’s an OKR, a measurable outcome, a projected ROI neatly packaged in a slide deck. I get it. But AI tooling doesn’t fit that mold right now — and forcing it into that mold is exactly why your adoption is stalled.
Instead, give your talented people access to AI tools and let them find their efficiencies organically. Trust the people you hired. They know where the friction is in their daily work better than any top-down strategy document ever will.
Things That Never Would Have Made It Into Sprint Planning
Here’s what I’ve seen happen when you get out of the way: teams start solving problems they’d given up on.
Not the problems on the roadmap. The other ones. The ones that never made it into sprint planning because the effort-to-impact ratio didn’t justify the sprint points. The automation that would save 30 minutes a day but would take two sprints to build. The internal tooling that everyone wished existed but nobody had bandwidth for.
With AI in the mix, the calculus changes. A two-sprint project becomes an afternoon. Something that seemed infeasible with available sprint points is suddenly possible in a fraction of the time. And your team already knows which problems to go after — they’ve been living with them.
The ROI Will Come
I won’t pretend I can draw you a straight line from “give everyone Claude access” to a specific dollar amount on a spreadsheet. That’s the honest answer, and if someone is promising you that line, be skeptical.
But here’s what I can tell you:
- Velocity increases. Teams ship faster when repetitive work gets automated away.
- Knowledge gaps shrink. Junior engineers level up faster when they have an AI pair programmer. Senior engineers tackle problems outside their specialty more confidently.
- Morale improves. People like solving problems, not grinding through boilerplate. Give them tools that eliminate the grind and they’ll find better problems to solve.
- Institutional knowledge surfaces. When teams use AI tools to document, template, and systematize their workflows, tribal knowledge stops living in one person’s head. I wrote about exactly this in From Tribal Knowledge to Living Procedures.
The ROI is real. It’s just not the kind you can predict in a planning meeting — it’s the kind you measure after the fact and realize the investment was obvious in hindsight.
Get Out of the Way
This is one of those rare moments where the right move is to get out of the way, even if nobody has explicitly shown you an A-to-Z plan for how your org “wins” with AI.
Put the guardrails in place — data handling policies, approved tools, clear guidance on what’s in-bounds and what isn’t. That part is your job.
Then step back and let your people work.
The orgs that figure this out now will compound their advantage. The ones still waiting for the perfect strategy will be writing that strategy with AI tools two years from now, wondering why they didn’t start sooner.
Drafted with assistance from Claude. The irony of using AI to argue for AI adoption is not lost on me.