Roadmaps that Resonate

TL;DR:

Traditional static roadmaps are obsolete in the age of AI. Success requires dynamic, audience-specific roadmap canvases that adapt to rapid changes. We introduce the concept of dynamic roadmap canvases as well as the ACT roadmap model ‘Act, Calibrate, Transform’ in favour of ‘Now, Next, Later’. ACT enables real-time execution, iterative learning, and bold long-term bets in response to the pace of change in agent-based tech.

Introduction

The AI explosion has turned “yearly roadmaps” into relics. In the same way email is going the way of the dodo, so is the static plan. Roadmaps matter most when they’re living, breathing battlefield maps. Vision? Stakeholder alignment? Non-negotiables. But in the age of autonomy, if your strategy isn’t as dynamic as the tech reshaping it, you’re already irrelevant.

I’ve recently seen lots of different posts related to roadmaps on LinkedIn; including one by a product management app ( Productboard ) talking about types of roadmaps, which sparked this one.

Primarily because I think the discussion and the angle is incorrect. It's not about the “types” of roadmaps in terms of kanban style or gantt style timelines.

The core things that roadmaps need to be are:

• audience focused
• highly dynamic

A boardroom slide ≠ an engineering team sprint view and no one wants a roadmap that is irrelevant 1 week after publishing.

Forcing all stakeholders to work around a monolithic roadmap is a recipe for misalignment. With confusion becoming the tax you pay for clinging to antiquated thinking.

Dynamic roadmap canvases, like that within Vee , demand audience-first ruthlessness. They should take advantage of the psychology of 'What's in it for me'. Engineers need execution velocity. Investors/execs want TAM expansion. Customers crave outcomes, not feature grids.

Build bespoke narratives, not checklists. Operationalise clarity. The future isn’t built by consensus—it’s seized by those who cut through noise with precision.

Let's do this!

1. Roadmaps are Key; Even When Everything’s Changing

Before we start, I think I talk for every product leader/executive, CEO, CFO ever.

Ever, ever.

When I say no one wants a sprint by sprint, scrum/agile gantt chart delivery drudgery style 'roadmap'. Don't take my word for it see:

Teresa Torres here, Kira Gorbunova here, and again here.

Reminder:

Product Management ≠ Project Management

Engineering delivery plans are an essential part of operational execution. But they've got close to nothing to do with the intent of a product roadmap. A great roadmap isn’t just a laundry list of features. It’s a narrative of value.

It says:

“Here’s how we’ll solve problems, ease JTBD, win markets, and redefine what’s possible.”

At the same time, it has to be flexible. Because when you’re operating in an environment driven by dynamic AI breakthroughs, trying to nail down the next 12 months in detail is like scheduling rocket launches in a hurricane.

Your mission; should you choose to accept it:

Build a roadmap that conveys multiple stakeholder needs without becoming a rigid, high-stakes document you can’t evolve.

2. Tailoring Your Roadmap to Different Audiences

Not all roadmaps are created equal.

The version you show to your engineering team should look different from the one you walk through with your board.

Below are the stakeholders that typically demand a seat at the roadmap table—and what they actually want:

NB: I like to share Claude artifacts so find it here. The following are a couple of examples of this might look like in reality:

Product's view

Functional descriptions with details on upvotes, value score, effort and whether it's now, next, later:

Executive's View

Grouped by XO Themes of Increasing ACV, Maximising NRR, and Increasing TAM/SOM. Note the descriptions are changed (via AI) to reflect the business outcome.

Rather than value scores and effort the quarter and the expected return is shown.

3. “Now, Next, Later” in the age of autonomy

You’ve probably used the “Now, Next, Later” framework created by Janna Bastow . In my opinion, it is one of the best ways to approach roadmaps irrespective of whether you're a start up or a large enterprise.

It's straightforward to understand and focuses teams on immediate priorities and near-term roadmap items, with the future left open-ended.

But....in the age of autonomy, when everything is flux owing to the pace of change in AI, the assumed waterfall nature of the model starts showing cracks.

You might be forced to pivot “later” into “now” because your biggest competitor just launched a game-changing AI solution. Or when “next” never arrives because the market shifted overnight.

In the Agentic AI era, features can become obsolete before you even ship them. 11x got to Series B and then OpenAI showed a 'little side project they're working on' which will, at best, cast a lot of doubt on the value of that as a viable business. In the same way Apple does when it adds native functionality or an app that negates the need for a 3rd party solution that built its entire business on a pain point of an ecosystem.

The classic framework just doesn’t necessarily give you the nuance or speed to adapt.

4. Introducing ACT : A New Roadmap Model

To handle the pace of change AI is unleashing, consider replacing “Now, Next, Later” with something more dynamic:

ACT - Act, Calibrate, Transform

1. A – Act

• Items in here are initiatives you’re actively shipping or iterating on right now. Think days/weeks (max) not months.
• Stakeholders can expect near-immediate impact and improvements.
• The emphasis is on rapid execution, with short time horizons to accommodate shifting priorities.

Act is the engine of real-time agility. In a world where AI reshapes markets overnight, static roadmaps are relics. The Act phase isn’t about incremental tweaks—it’s about relentless execution at the speed of now. By compressing cycles to days or weeks, teams ship value fast, turning volatility into velocity. This isn’t “move fast and break things”; it’s move fast and build things that matter, with every release informed by the latest competitive, technical, and user realities. The dynamic nature of Act ensures organisations don’t just survive chaos; they weaponise it.

2. C – Calibrate

• Projects in this bucket are still in motion but likely need data-driven refinement or user feedback to move forward.
• You might be prototyping a new AI model, running a pilot, or testing a high-risk feature.
• Timelines are flexible—if you learn something game-changing, you can pivot these into Act or freeze them for later.

Calibrate is where market intelligence becomes action. We’ve seen companies die from clinging to assumptions. The Calibrate phase codifies what winners already know: feedback isn’t a retrospective—it’s oxygen. Here, AI-powered pilots, prototypes, and experiments aren’t vanity projects; they’re live wires into the market’s pulse. When OpenAI drops a model or Apple swallows your niche, Calibrate lets you pivot before the disruption becomes existential. This is how you turn threats into R&D—by stress-testing hypotheses against real-time data, not deckware.

3. T – Transform

• These are your moonshots, your big bets, your 10x leaps.
• They might be AI-driven expansions into entirely new domains—or next-gen solutions that haven’t proven feasibility yet.
• The goal is to inspire but not over-promise. You track them, measure early signals, and pull them into Calibrate when the time is right.

Transform is how you future-proof. Vision without mechanism is hallucination. The Transform phase forces teams to institutionalise audacity and a strategic lens: those 10x leaps, ecosystem redefinitions, and bets on AI’s uncharted frontiers. It’s the hard work of future-building—tracking weak signals (a regulatory shift, an open-source breakthrough), nurturing moonshots with discipline, and pulling them into the calibrate forge when the moment ignites. Transform isn’t optional in the age of agentic AI; it’s how you ensure today’s Netflix doesn’t become tomorrow’s Blockbuster.

It also turns out that this line of thought is not new. Having discussed this with a colleague Alix Cunnell she highlighted that it is effectively a reimagining of McKinsey & Company 's three horizons model and something Steve Blank had highlighted 5 years ago now. Technology is now evolving 10-fold faster than it was back in 2019 so the ACT roadmap model recognises that in fast-evolving markets, you need multiple feedback loops, not just one linear waterfall progression from now to next to later.

With AI breakthroughs dropping by the day, “Act, Calibrate, Transform” keeps the focus on immediate impact, iterative experimentation, and long-range disruptive potential—all at once.

5. Closing Thoughts: Navigating the Future

The ACT framework rejects the linear, tick-the-box waterfall. In markets moving at AI-speed, survival demands real-time feedback loops—not a stale march from “now” to “later.” Act, Calibrate, Transform isn’t a roadmap; it’s a live-fire drill for balancing immediate execution, ruthless market learning, and 10x leaps. When AI breakthroughs drop daily, you don’t “plan”—you pressure-test.

Your roadmap isn’t a PDF. It’s a breathing, battle-ready playbook. The ACT model forces teams to operate on the assumption that disruption will happen: shipping today’s edge (Act), stress-testing tomorrow’s bets (Calibrate), and future-building in the face of existential unknowns (Transform). This isn’t agility—it’s antifragility.

Deliver value or die. Incremental wins? Ship them. Hard pivots? Make them. Moonshots? Track them. A roadmap worth its code supercharges your ability to seize chaos, not cower from it.

Build fast, adapt faster, and lean into the storm. In the Agentic AI era, the winners won’t follow maps—they’ll rewrite them. Do that, and your roadmap stops being a document. It becomes your category-defining weapon.

Follow me John McMahon and Vee for more fresh takes, new models and counter-intuitive thought flows.