Your AI Roadmap Is Already Obsolete. Here's Why That's a Finance Leadership Problem.
Your AI Roadmap Is Already Obsolete. Here's Why That's a Finance Leadership Problem.
A few weeks ago, a colleague and I were mapping out an AI project — roadmap, sequencing, what was coming next. By the time we circled back, one item had already been superseded by a capability that shipped in the gap between our conversations. The roadmap wasn't wrong when we wrote it. It was just no longer the right question.
Finance leaders across different sectors are telling me variations of the same story. The AI roadmap that seemed ambitious in January looks conservative in June. The tool shortlisted in Q1 has been leapfrogged by something released in Q2. The integration that required custom development last year is now a native feature. The pace isn't just fast — it's fast enough to make traditional roadmap thinking actively counterproductive. That's a finance leadership problem: not because finance leaders own AI strategy, but because the frameworks finance brings to planning and governance are exactly what needs to adapt.
|
Weekly
Pace of meaningful AI model and product updates in 2026 — frontier capability shifts that would have been annual events are now arriving on a rolling basis, faster than any traditional planning cycle was designed for
|
Moving frontier
Benchmark leaders are changing quickly enough that tool choices made six months ago deserve active review — the roadmap item that was "next quarter" may already be a native feature today
|
|
New bottleneck
AI has removed many execution bottlenecks — but the constraint has shifted. In 2026, the scarce resource isn't capability. It's the human time and attention needed to direct it well
|
Slow adoption
Australian economists and productivity commentators consistently point to slow technology adoption — not capability access — as the reason AI's productivity gains are taking time to show up in national figures
|
The Roadmap Problem Is a Planning Architecture Problem
The traditional AI roadmap works like a capital project plan: define the requirements, select the tools, sequence the implementation, allocate budget, execute. It assumes that the landscape is stable enough that a plan made in one quarter is still relevant in the next. For most technology projects, that assumption holds reasonably well.
For AI in 2026, it doesn't. Not because the underlying approach is wrong, but because the rate of capability change means that the specific tools and features on a roadmap have a shorter useful life than the planning cycle that produced it. A roadmap item that required a custom integration six months ago may now be a standard feature. A model that represented best-in-class performance when the roadmap was written may have been outperformed twice since then.
Finance leaders involved in AI budgeting will recognise this immediately. The business case written in November, approved in February, funded from July — it's working with AI capability assumptions that are already stale before the project starts. That's not a failure of the team; it's a structural mismatch between planning cycles and the pace of the technology.
What Finance Brings to This Problem
Finance has tools well-suited to managing this kind of uncertainty — they're just not typically applied to AI planning. Rolling forecasts hold a budget envelope at the capability level ("we're investing in AI-assisted reporting") while reviewing specific tool choices quarterly. Stage-gate funding releases budget in tranches contingent on defined outcomes, building in reassessment before the next commitment. These aren't new ideas — finance uses them for capital projects regularly. The insight is that AI investment warrants the same approach: high-uncertainty environments need adaptive planning architecture, not full-scope upfront commitments.
The Governance Question: Who Owns AI Currency?
One of the practical challenges organisations face is that nobody has explicit ownership of "staying current on AI capability." Technology teams track new releases in their domain. Strategy teams look at competitive positioning. Finance teams manage the budget. But the question of whether the AI tools and approaches the organisation is investing in are still the right choices — relative to what's available now, not what was available when the decision was made — often falls through the gaps.
Finance is well-placed to own this — not by becoming AI experts, but by building regular AI investment reviews into the budget cycle. A quarterly review asking: are our AI tools still best-fit? Have capabilities emerged that would shift our priorities? What have we stopped using, and why? — doesn't require technical expertise. It requires the same structured questioning finance applies to any investment portfolio. Organisations handling this well tend to have a small designated group — sometimes just one or two people — whose job includes monitoring AI capability and translating it into business implications. Not a full AI team: an early-warning function that no annual roadmap can replicate.
From Roadmap Thinking to Direction-Setting
The shift worth making isn't from "do roadmaps" to "don't do roadmaps." Planning, sequencing, and budget approval are still necessary. The shift is in how tightly the roadmap is held — and what role finance plays in maintaining the flexibility to update it.
A roadmap in a fast-moving AI environment should define direction, not dictate specifics — describing where the organisation needs to get to and what capabilities it needs, without locking in specific tools and integrations beyond one or two planning cycles. Finance leaders who can reframe the conversation from "here's what we're building and when" to "here's where we're going and how we'll adapt" are providing strategic value that AI-generated analysis can't replace. That's a distinctly human finance capability, and it's the one that matters most right now.
The New Bottleneck: It's Not the AI Anymore
There's an observation I keep seeing in conversations about AI adoption in 2026, and it matches my own experience exactly: the bottleneck has shifted. For years, the constraint was capability — AI couldn't do the thing well enough, reliably enough, or at the right cost. That constraint has largely lifted. The tools exist. The capability is real. The new bottleneck is human time and attention.
I can speak to this directly. The list of things I want to build with AI is longer than it's ever been. The capability to build them is genuinely within reach. But building them well still requires directed human effort — framing the problem correctly, reviewing output, testing against real scenarios, iterating. That takes time which is, in practice, finite and heavily competed for across multiple responsibilities. The gap between what's possible and what there's genuine capacity to do properly is a real constraint — and one that most AI roadmaps don't account for explicitly.
The practical implication: when scoping AI initiatives, the binding constraint to model isn't tool cost or technical feasibility. It's leadership bandwidth. How much directed human time does this initiative require to implement properly, govern responsibly, and improve iteratively? Reframing the AI planning conversation from "what can we build?" to "what can we sustainably direct given our current capacity?" is exactly the kind of grounding that prevents organisations from accumulating half-implemented AI projects nobody has the bandwidth to maintain.
PFL works with finance leaders on AI investment strategy — from structuring business cases that hold up in fast-moving environments to building governance frameworks that keep AI decision-making accountable without creating paralysis. If your AI roadmap is starting to feel like it's always already out of date, we should talk.
Talk to PFL →
Comments
Post a Comment