Strategy Meets Autonomy
Updated: Jul 26
How to Align People, Data & Agents
We’re moving from the era of “digital transformation” into a new phase: digital autonomy. It’s not happening overnight—and that’s normal. We tend to rush into new trends and only later realise they aren’t magic solutions.
Recent stories about AI "agents" make that clear. Klarna, for example, put major effort into building an AI-driven workforce culture, but a year later, the promised gains in efficiency and customer satisfaction reportedly didn’t fully materialise.
Machine learning (a subset of AI) has been widely used for years, but many now see it shifting from making recommendations to making decisions. That’s why there’s so much excitement: Agents are no longer just support tools—they’re starting to become central to how businesses operate.
If you’ve seen real-world examples where agents have successfully replaced human roles, this is a good time to share them.
Why this matters
The real point of this article is simple: we need to rethink our strategies—and, even more importantly, our operating models.
If your organisation still runs on traditional hierarchies and human-only decision processes, you’re already behind. The biggest challenge isn’t just what AI can do; it’s how to align that capability with human expectations around fairness and accountability.
So the key question becomes: how do we align agents and humans around a common set of business goals?
This article lays out a practical framework to reshape your data, business, and operating models for a world where humans and agents work together. And, importantly, it avoids the hype and sensationalism around “agentic AI.”
Let’s bridge the gap
Many organisations still treat business strategy and data strategy as two separate tracks: data teams build insights and models, while business teams chase growth. This framework is designed to close that gap.
Because agents operate in real time and across silos, they need context. If your business goals aren’t embedded in your data and AI foundations, Agents will either make the wrong moves—or simply won’t make a meaningful impact.
Given how long we’ve talked about breaking down silos, it’s clear we don’t need two strategies. We need one outcome-driven strategy, strengthened by data, and executed jointly by humans and agents.
A practical framework to align strategy, data, and the operating model
I’ve been working in data strategy since 2002, and over the years, I’ve developed frameworks and canvases (including the data strategy canvas) to help leadership teams turn principles into execution.
This framework has five steps, based on one core idea: Agents aren’t outside the business—they’re part of how the business runs.
Define the business outcomes Agents can drive. It’s not enough to say you want to “improve efficiency” or “enhance the customer experience.” You need clear SMART goals—and you need to spell out which decisions and actions Agents should take.
Two key questions for leadership and the executive sponsor:
If these questions are hard to answer, your organisation probably isn’t ready for agents yet—because agents need direction, and that direction has to come from leadership.
Where are decisions slow, repetitive, or inconsistent today?
Which actions are you genuinely comfortable delegating to a machine?
Rebuild the operating model for human–agent collaboration. This is a major shift. Operating models connect strategy to execution—business to IT, company to customer—but they’re often treated as an afterthought.
Most operating models today are designed for humans: approvals, meetings, and oversight. But agents are built to act quickly, not to sit in process bottlenecks. You need an operating model where:
This kind of collaboration has to be designed on purpose. If it isn’t, you’ll get confusion, resistance, or (worse) blind trust in bad decisions.
Agents have defined roles (like humans) with clear mandates and boundaries.
Humans and agents work in loops, not handoffs—more like product teams where agents do the work and humans review, guide, or refine.
Decision rights are shared or hybrid, with clear accountability for who owns which actions.
Anchor everything in use cases. Use cases are essential. This isn’t about writing long strategy documents that no one reads—it’s about choosing a small number of high-impact use cases that connect business outcomes, data, technology, and operations to real value.
Strong use cases should:
This is where strategy becomes real—and where you move past “tech stacks built for ego.”
Be led by the business—not just tech or data teams.
Have clear value metrics (for example, lower service costs or higher customer retention).
Require changes in both technology and day-to-day processes.
Build a lightweight, responsive data infrastructure. Agents need data—but not all the data. They need the right data: purpose-fit, context-rich, real-time, and available across silos.
Data teams need to shift from “building platforms” to enabling decisions and action. Speed, relevance, and quality matter more than perfection. A minimum viable data foundation that supports action beats a “perfect” system that slows everything down.
And the focus shouldn’t be only on architecture—it should be on how decisions and actions happen: who acts, where, and when.
Redefine success metrics. Measurement has to move away from purely tech- or data-centric reporting and beyond traditional ROI models tied to headcount reduction.
Instead, measure what directly supports your goals, including:
If executives can’t see (and measure) the value agents create, they’ll either underinvest—or invest in the wrong things.
Decision velocity: how quickly good decisions get made.
Human lift: how much time and mental load teams get back.
Process compression: how many unnecessary steps do agents remove?
This is about redesigning the business—not upgrading technology
A lot of organisations are still figuring this out, and that’s understandable. But introducing AI without redesigning how the business actually operates only scratches the surface of what’s required.
This shift demands alignment across strategy, data, humans, and agents—so decisions are clear, value creation is measurable, and workflows actually work.
If leadership doesn’t shape that alignment, the technology will end up dictating the terms. And history suggests that it rarely ends well.
To close, here are three questions worth reflecting on:
Are your business strategy and data strategy truly aligned?
Does your operating model actually support human–agent collaboration?
Or are you just layering AI onto an ageing system that can’t keep up?
The time to answer those questions is now.
