🧩Four Modes of Enterprise AI Investment: Adapt | Augment | Automate | Transform

CAIO Enterprise AI Portfolio series, Part 1

This strategy note is Part 1 in a subscriber-only series outlining the Enterprise AI Portfolio Management framework for CAIOs that I use in Memia’s AI Strategy client engagements. As always, please reach out with comments, questions and enquiries.

Reconciling AI / real-world economies

Increasingly it feels like we exist in a dual-economy world…

On the one hand, I spend my days immersed in the online AI hype monster, drinking from the continuous firehose of new frontier models, AI lab product announcements, multi-billion-dollar investment rounds, AI CEOs promising AGI *yesterday*, LinkedIn bros bragging about their 24/7 Claude Code vibe coding exploits and agent swarms… If you let yourself believe in the hyperbole from ingesting this never-ending stream, then AI is always a 100X productivity booster, whole lifetimes of work can now being achieved in a few hours and — crucially — endless supplies of VC money are available for even the most half-baked AI proof-of-concepts...

But then when I engage with “real-world” enterprises outside the online AI industry bubble, pretty quickly I get jerked back down the gravity well. Suddenly robust scepticism, value measurement (how’s the ROI from that company-wide Copilot rollout going?), security risk aversion, concerns about cost inflation, HR impacts and skills training are instead at the fore. Plus, in many organisations I have worked with, there remains entrenched doubt at senior levels that AI will have any net disruptive or enabling effect on the bottom line for the foreseeable… combined with a tendency to over-index on headline-grabbing industry reports that most (eg 95%) AI pilots fail (more about this particular dynamic in another post…)

Bridging the gap between these two worlds is the daunting task facing CAIOs, CIOs and CDOs today: how to translate the fast-moving, hyped-up, accelerating world of frontier AI into tangible enterprise value terms so that time-challenged, innately sceptical, non-technical decision makers can gain enough confidence to greenlight direction and budgets.

Communicating the AI investment case

Communicating the enterprise AI investment case is complex on many fronts, precisely because there are so many conflicting signals. On the one hand, it’s hard enough just batting away the anecdotally high failure rate of most AI initiatives. But conversely, CAIOs must also deal with the opposite trendline: that agentic capabilities of AI seem to be doubling every 7 months or so…

METR

… which makes it even harder to build a case for investing now rather than just waiting for the next model:

METR

(Case in point: why invest in OpenAI o4 in April 2025 when Claude Opus 4.5 blows it out of the water just 7 months later? Looked at in isolation, the retooling / rework / retraining costs alone would seem to negate the original project’s ROI.)

And be in no doubt: things are moving faster every day, while most organisations demonstrably don’t have the capabilities to adjust to this accelerating pace. As AI Innovisory CAIO Peter Mangin put it succinctly on LinkedIn this week:

“AI moves fast enough now that almost any team can ship something impressive in days. That part isn’t the differentiator anymore.

What decides whether it survives is what happens next. When more people touch it. When real data shows up. When it quietly starts influencing decisions.

That’s where most AI work … stalls. Not because the tech doesn’t work, but because ownership, boundaries, and risk were never designed in.

The teams that scale don’t slow down. They add just enough structure to keep momentum without losing trust.

You don’t need a six-month AI strategy.

You do need a way to turn experiments into something your organisation can actually rely on.

Speed gets attention.
Maturity keeps it.”

I agree 100% here: enterprise AI investment needs BOTH speed AND operational maturity. To achieve this, organisations will need an agile, measurement-focused portfolio framework that enables both. To all intents and purposes, the AI strategy IS the portfolio.

Four modes of enterprise AI investment:

Over the last few years I’ve worked as an advisor with diverse clients helping to envision, articulate, shape and manage AI Strategy. Over these engagements, I’ve evolved what I’ll call the Memia Enterprise AI Portfolio Framework¹ through a series of iterations, helping to enable CAIOs and their stakeholders to manage AI innovation and measurement at pace and scale.

This series of strategy notes in 2026 is intended to summarise the fundamental concepts of the framework, where it’s got to so far, what tooling support exists² and some worked examples of how to adopt and implement the framework. (Everything is released under CC BY licence so feel free to take this,fork it and run with it by yourself… at some point soon I’ll open-source the software tooling when I have time…!)

The first step when building any portfolio is to articulate the various time horizons, investment theses and measures of success or failure. From there, the portfolio can be populated, managed and directed over time towards a strategic optimum balance.

To start, in my experience most enterprise AI strategy can be boiled down to considering two simple axes:

Axis 1: Strategic intent, Protect ←→Grow

• Is the intention to use AI to grow the business or protect what we’ve already got?

Axis 2: Unit of change, People ←→ Processes

• When investing in AI to drive change, is the primary focus to enable people to do more, OR make processes more efficient so we need less people? (…OR, transformatively: both?)

Everyone loves a good 2X2 quadrant, so composing these axes yields four distinct motivations for enterprises to invest in AI:

MEMIA (CC BY)

Let’s dive into each of these in more detail: