Stop averaging your AI maturity score

Maturity models hand leaders one number. The companies that get value from AI choose a profile instead deep in the value streams that pay, exploratory where regulation is still moving, with one set of governance rules underneath.

Why does a single AI maturity score leave your next budget decision exactly where it was?

A single AI maturity score leaves your budget decision where it was because it describes no part of your business accurately. Your fraud team runs models in production while marketing tests prompts. Average the two and you get a middle number: the strong area reads as ordinary, and the weak area reads as acceptable. Neither team gets a clear mandate, and neither budget line has a defensible rationale.

The board question is real. Where does your company stand on AI? A five-level answer fits on a slide. What it cannot tell you is which function to fund next, which one to hold back, or which regulatory exposure is sitting in your score without anyone tracking it.

The thesis here is simple: put each function at the maturity level its work justifies, make the gaps between functions a deliberate choice your leadership team signs, and apply one governance baseline to all of them. Unevenness is often the correct target state, not a problem to fix.

What do the Gartner and MIT CISR AI maturity models get right and what do they assume?

The established AI maturity models produce real evidence. Gartner’s 2025 survey found that 45% of high-maturity organizations kept AI projects in production for three years or more, against 20% of low-maturity organizations. MIT CISR’s four-stage model links maturity directly to financial performance: enterprises in stages 1 and 2 performed below their industry average, while those in stages 3 and 4 performed well above it. The largest single financial gain appeared between stages 2 and 3. Both models are worth using.

The assumption worth questioning is the one both models share: they draw maturity as a line and place the whole organization on one point of it. Academic reviews of AI maturity frameworks flag that linear, organization-wide progression assumption as the part that breaks first when you try to act on it. It is not wrong to measure this way. It is incomplete when you need to decide where to spend.

Many consultancies already advise measuring maturity by function rather than by company. Ceiba’s Augmented Organization Model takes the next step: it treats the variance between functions as a target your strategy sets, not a gap to close uniformly. The profile you choose is a budget plan.

Where does AI actually create value and does your maturity depth match that?

AI’s potential value concentrates in a small number of core functions. BCG’s 2025 survey of 1,250 executives placed roughly 70% of AI’s value in functions like R&D, sales and marketing, manufacturing, and pricing. Only 5% of companies qualified as ‘future-built,’ and those companies share one behavior: they redesign core workflows first. They have deployed most of their AI initiatives, while laggards have deployed few.

The maturity consequence is direct. If two or three value streams hold most of your value, building depth there first returns more than lifting ten support functions one level each. Taking one core stream from stage 2 to stage 3 in MIT’s terms is a better investment than a broad, shallow lift across the organization.

McKinsey’s State of AI 2025 reports that high performers use AI in more business functions than their peers, which can read as an argument for breadth. The order of operations matters. High performers redesign workflows end to end in their core streams and expand from there. Breadth follows depth; it does not replace it.

The cost argument runs the same direction. A uniform maturity target means paying for advanced tooling and change programs in functions that cannot return the investment. An uneven profile with AI spend matched to the value each function can produce is a CFO argument, not just a technology choice.

When is Explorer the right target and what stays the same across all functions?

Explorer is the right target for a function when it faces regulatory exposure that makes production systems premature, when its workflows carry too little value density to justify agentic redesign, or when the organization’s change capacity is already committed to higher-priority streams. Holding a function at Explorer is a decision, not a failure.

Regulatory exposure is the clearest case. The EU AI Act classifies AI used for hiring, promotion, performance evaluation, and worker monitoring as high-risk. The EU Digital Omnibus moved those obligations to December 2027, but it did not remove them. US states continue rewriting their rules on AI in employment decisions. While the rules are still in motion, keeping HR at Explorer means your team can test and learn without building production systems it may have to rebuild when the rules settle. This is a risk management call, not a comment on what AI can do in HR or on anyone’s role.

Low value density is the second case. Some workflows run a few times a quarter or carry limited financial weight. An AI assistant handles them without agentic redesign that costs more than it returns. Change capacity is finite; spending it here is a trade-off against spending it in a core stream.

One rule holds regardless of target level. Data stewardship, access controls, security policy, and basic AI literacy apply to every function, at any maturity stage. Your HR team at Explorer follows the same data rules as your software engineering team at Exponential. The baseline is not optional; the ambition varies.

Ceiba applies the same logic at workload scale through the AI Autonomy Spectrum. Senior engineers review densely where the stakes are high and lightly where they are low. The maturity profile extends that calibration from a single workload to a whole function.

How does the MOA’s maturity assessment produce a profile instead of a score?

The MOA’s maturity assessment scores each of its four axes: people, operations, governance, and tools at three levels (Explorer, Intermediate, Exponential) for each area of the business separately, not for the company as a whole. The result is a profile, not a score. Ceiba’s consultants then set a target profile with your leadership team against your strategy, not against a generic benchmark.

MOA Axis	Explorer	Intermediate	Exponential
People	AI literacy policy; no active program	Champions network active; role-based training	Authors publishing agents; AI embedded in career paths
Operations	Manual workflows; isolated experiments	Core value streams redesigned end to end	Continuous agent improvement; metrics tied to business outcomes
Governance	Baseline data and access controls	Risk tiering; autonomy spectrum applied per workload	Full audit trail; Sextante + Astrolabio operating
Tools	Productivity tools and assistants	Spec-driven development in priority streams	Model-independent architecture; AI FinOps cadence

Maturity is a sequencing input, not a destination. Sextante, one of the MOA’s instruments, ranks incoming initiatives by value, risk, and current maturity before anyone commits budget. A team scopes an initiative in an Explorer function differently from one in an Exponential function. Astrolabio checks afterward whether the promised value appeared. Both instruments stay with your organization when the engagement ends.

Ceiba’s own profile is the proof of concept. Ceiba took software delivery, its most exposed value stream, deep first. The result is the Ceiba Method, running on approximately 116 active projects. Other functions of the company moved at their own pace. The unevenness was a choice.

The nearshore mechanism makes the profile operational. Senior Latin American engineers and consultants in US time zones embed in the function you choose to move first. They join the planning sessions where your leaders set the target profile, which means the team doing the work is in the room when the priorities are set.

Grid showing AI maturity levels (Explorer, Intermediate, Exponential) by MOA axis and business function. Software Engineering shows Exponential across all axes. HR shows Explorer across all axes with a note about EU AI Act high-risk classification for employment AI. Sales and Finance show mixed profiles across people, operations, governance, and tools axes.

Which function moves first and which one holds at Explorer with a reason next to it?

The question every maturity assessment should leave you with is not ‘what is our score?’ It is ‘which function moves first, and which one holds at Explorer with a written reason next to it?’ The first question has a budget attached. The second has a governance decision attached. Both belong to your leadership team, signed.

Bring the maturity diagnostics you already paid for and your initiative portfolio.

Ceiba’s consultants will break the single score into a per-axis profile, set a target with your leadership team, and outline a three-month experiment in the function that should move first.