A few weeks ago, I sat down with a Fortune 500 chief architect and a good friend of mine to talk about what it actually takes to get an enterprise ready for agentic AI. It’s a conversation I’ve been having in different forms all year, most recently at the inaugural US AI Congress in Washington, D.C., where more than 50 members of the 682-person-strong Chief Architect Network of Fortune 500 architecture executives showed up to compare notes on exactly this problem.
Most organizations are deploying agentic AI onto systems that are heterogeneous, brittle in places, and held together by years of decisions made for reasons that made sense at the time. The instinct is to sprinkle agentic AI on top and hope it helps. It doesn’t. That's what I invited Grant Ecker (Founder of Chief Architect Network) to join me on “What’s the BUZZ?” and share what can enterprise architects can do to lead the necessary progression. Here’s what we talked about…
Business Capability Mapping Beats Sprinkling AI on Top
The early wins in agentic AI almost always look the same: a single use case, an out-of-the-box tool, a faster turnaround on one task. Those wins are real, but they run out quickly, because they rest on a flawed assumption. The assumption is that the business will keep operating exactly as it always has, and AI will just help it do that a little faster.
The better assumption, the one my guest kept coming back to, is that AI is going to transform how the business operates, which means leaders need to rethink horizontal capabilities rather than bolt AI onto vertical projects one at a time. The tool for that is business capability mapping: take the company strategy, decompose it into the capabilities the business has to be great at to deliver on it, and then honestly assess where the gaps are. Some of those capabilities are obviously exciting, like product innovation. Others are unglamorous, like order management or campaign execution, and some aren’t technology capabilities at all. Map all of that against every strategic priority, and what you get is a heat map of where transformation actually pays off, instead of a list of wherever a vendor happened to pitch you first.
This is also where the “if it ain’t broke, don’t fix it” instinct has to give way to something closer to the opposite: if a system is fixed in the middle of a transformation and it’s sitting on a capability that matters to the strategy, that’s exactly the system to open up and rebuild. The test is whether fixing it moves a capability that the business actually depends on.
Get your copy: The HUMAN Agentic AI Edge
Organizations are racing to deploy Agentic AI, yet few are ready for the risks that emerge when employees use AI without structure, standards, or oversight.
The HUMAN Agentic AI Edge offers leaders a practical blueprint for building accountable AI-ready teams that consistently produce high-quality results. Drawing on real-world knowledge and insights from interviews with more than 50 AI leaders and experts, Andreas Welsch shows how to combine human judgment with Agentic AI capabilities to achieve the performance many organizations expect but rarely deliver. This book prepares you to shape the next generation of AI-ready teams delivering high-quality results with high accountability.
Every architecture leader I talk to wants to be in the room when the business sets strategy. Very few get there by asking to be invited. But you don’t get to drive the car everyone admires by copying what the driver wears. You earn it the same way IT leaders earn credibility with the business by first earning trust inside IT.
That order matters more than it sounds like it should. If you build relationships with business stakeholders before you’ve built trust with your own IT organization, the first thing the business will tell you is whatever their IT partner has been quietly warning them about for months. Agree with it without knowing the full picture, and you’ve undermined the people who actually own that relationship, possibly for good. Build the IT trust first, find the shared outcomes, and the business relationships follow naturally instead of working against you.
There’s a skills shift underneath this, too. Architecture has trained people to be T-shaped: deep in one area, broad enough to connect to the rest. That’s no longer sufficient on its own. What’s needed now is closer to pi-shaped, where technical depth is paired with real business fluency, the ability to understand what the business consumes, what it needs, and how to communicate in those terms. Depth without that business layer keeps you out of the room, no matter how good the depth is.
AI FinOps: Forecasting Cost Before It Forecasts You
As AI labs move from flat per-seat subscriptions to consumption-based, token-driven pricing, the bills are starting to arrive, and they’re catching leaders off guard. The fix requires the same discipline utilities have always needed: know who’s consuming what, forecast it, and route the cost to the right part of the business before the surprise hits the P&L. AI FinOps is becoming a standing agenda item for architecture leaders for exactly this reason, and it deserves the same seriousness as any other capital planning exercise, tied to the ROI the spend is supposed to produce rather than the spend itself.
The other piece of cost discipline is time. Because this space moves so fast, by the time a best practice is written up and published, it’s already six to nine months old. If you’re trying to stay ahead of the curve rather than catch up to it, published content isn’t where that edge comes from. It comes from a trusted peer in a non-competing industry who’s a step or two further into the same problem and is willing to compare notes. That’s a genuinely underused lever, and it’s worth building deliberately rather than hoping it happens at the next conference
Summary
Getting your enterprise architecture ready for agentic AI starts with mapping the business capabilities that actually matter to the strategy, earning the trust that gets you into the room where those capabilities get decided, and building the cost discipline to fund the work without a budget surprise six months in. Agentic AI does not fix a broken architecture. It just makes the cost of not fixing it more visible, faster.
So here’s the question worth sitting with if you lead architecture, IT, or a business function trying to get more serious about agentic AI: are you still looking for the next use case to sprinkle AI onto, or have you mapped the handful of capabilities where fixing the underlying architecture would actually move your strategy forward?
Equip your team with the knowledge and skills to leverage Agenti AI effectively. Book a consultation or workshop to accelerate your company’s AI adoption.
Join my bi-weekly live stream and podcast for leaders and hands-on practitioners. Each episode features a different guest who shares their AI journey and actionable insights. Learn from your peers how you can lead artificial intelligence, generative AI, agentic AI, and automation in business with confidence.
"Business capability mapping beats sprinkling AI on top" is the same conclusion I keep reaching from a different angle: most orgs treat agentic AI as a faster way to run the same vertical project, when the real leverage sits in the horizontal capability underneath it.
The AI FinOps point is the sharper half, though. Token-driven, consumption-based pricing means the bill scales with usage in a way flat per-seat subscriptions never did, and most architecture teams are still treating that as an accounting surprise instead of a capital-planning line tied to the ROI the capability is supposed to produce.
Genuinely curious how your 682 chief architects are deciding which unglamorous capability, order management, campaign execution, gets rebuilt first once the heat map says it matters, versus which stays untouched because it isn't broken yet. Happy to connect, let's talk more about it.
"Business capability mapping beats sprinkling AI on top" is the same conclusion I keep reaching from a different angle: most orgs treat agentic AI as a faster way to run the same vertical project, when the real leverage sits in the horizontal capability underneath it.
The AI FinOps point is the sharper half, though. Token-driven, consumption-based pricing means the bill scales with usage in a way flat per-seat subscriptions never did, and most architecture teams are still treating that as an accounting surprise instead of a capital-planning line tied to the ROI the capability is supposed to produce.
Genuinely curious how your 682 chief architects are deciding which unglamorous capability, order management, campaign execution, gets rebuilt first once the heat map says it matters, versus which stays untouched because it isn't broken yet. Happy to connect, let's talk more about it.