Building the capability

Whether people are actually equipped to use it, and who paid for that.

67documents on this topic
26organizations represented
5issues named
14sourced citations
0sourced statistics

The state of it

One of 5 topics within Change leadership & adoption.

67 documents from 26 organizations address whether people are equipped to use what has been bought. It is the largest topic in this theme and the most conventional: the recommendations are training, literacy programmes and structured learning, and they are broadly the same everywhere.

What distinguishes the stronger material is treating fluency as something built and maintained rather than delivered once. BCG's board work is the clearest example - structured learning agendas, annual retreats, quarterly immersions, refreshed often and connected to outcomes - and it applies the same standard to directors that everyone else applies to staff.

The gap across the set is funding. Capability is named as the largest share of the required investment and is rarely costed as a standing line.

The issues, by agreement

How many independent organizations name each issue as a problem. An issue is only as real as the number of separate publishers that identify it, so the count is the ranking. Bars are organizations, not documents. Where the count reads ours, no publisher here states the issue and the analysis is our own.

Who takes which position

The chart above counts positions; this shows whose they are. Read down a column for what one organization holds across the whole topic, and across a row for who lines up on one issue. Where a cell carries more than one position, the strongest is shown and the rest are in the tooltip.

Ddisputes it Qqualifies it Nnames it as a problem Pproposes a fix
Building the capability: 5 issues against the 6 organizations cited on them. The number under each name is how many of these issues it is cited on.
Issue Carnegie Mellon SEI · 4 Deloitte · 2 McKinsey & Company · 2 Accenture · 1 Boston Consulting Group · 1 KPMG · 1
The AI plan and the people plan are written by different people, at different speeds N N P N · ·
The strategy is written, unimplemented, and treated as though it were done N Q P · · ·
Fluency is delivered as a programme rather than maintained as a capability · · · · N P
Every level is treated as a rung to climb, when a lower one is sometimes the right answer N · · · · ·
What a maturity average can hide P · · · · ·

A dot means this organization is not cited on that issue. It does not mean they are silent on it: an organization is cited where its document takes a position we could locate, and the absence of a citation is the absence of a finding, not a finding of absence. Who is represented lists everyone working on this topic, including those not cited above.

Where they disagree

No contradictions recorded on this topic yet.

The issues in full

Each issue carries the organizations that name it, the numbers behind it, and the remedies proposed - with the concrete steps under each. Every citation points at a section of a named document, so any count here can be checked.

Issue 013 organizations name it2026 evidence

The AI plan and the people plan are written by different people, at different speeds

Technology adoption and workforce capability advance on separate tracks with separate owners. Two thirds of organizations say the two plans are not aligned, which shows up as capability arriving after the system that needed it.

Accenture surveyed 1,320 C-suite executives and 4,560 employees across 20 industries and 12 countries in August and September 2025. Only a third said their talent strategy is fully aligned to their technology and AI strategy. The firms where it is aligned are a minority - 18% of those studied - and they report materially better outcomes on culture, employee experience and adaptability, plus an 11% uplift in innovation-related skills. The mechanism is unglamorous: functions set their own priorities in silos, so upskilling and deployment move at different rates and the gap between them is where adoption stalls. This is not a training budget problem. It is two plans that were never required to reconcile.

How to fix it — 1 approach, 3 steps

Make it one plan with two columns

Every deployment line gets a capability line beside it, with the same date and a named owner. If the capability column is empty, the deployment is not ready.

Done when The deployment and capability roadmaps sit on one page item by item, each pair has a single joint owner, and no go-live is scheduled against an unstaffed capability line.

  1. Put the deployment roadmap and the capability roadmap on one page, item by item.0-30 daysCHRO
  2. Give each pair a single joint owner so neither can be signed off alone.30-90 daysCEO
  3. Block any go-live whose capability line is unstaffed.90-180 daysCOO
The evidence — 4 documents
OrganizationDocumentPosition
AccentureConsultancyReinventing talent for the AI eraOur reading Reports only a third of organizations with a talent strategy fully aligned to their technology and AI strategy, and identifies the aligned minority at 18% of those studied, with an 11% uplift in innovation-related skills.Clarity; research snapshotnames it
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading Names insufficient knowledgeable staff among the three leading brakes on adoption, alongside data and legacy integration, and puts workforce literacy at the first maturity level - so the people side is a stated constraint rather than a soft one.Top factors negatively impacting adoptionnames it
DeloitteConsultancy · March 2026Deloitte's AgenticAdopt frameworkOur reading Reports fewer than half of organizations having begun to change how they upskill and reskill, which is the same misalignment observed as inaction rather than as intent.Building a culture of continuous learningnames it
McKinsey & CompanyConsultancy · August 2026Agentic AI change leadershipOur reading Proposes designing the change around what people actually fear, which only works if the people plan is drafted alongside the technology one rather than after it.Leaders who name and design for the fearsproposes a fix

Issue 021 organization name it1 qualifies it2026 evidence

The strategy is written, unimplemented, and treated as though it were done

Most organizations have an AI strategy on paper that has not been carried through. Because the document exists, the gap reads as execution lag rather than as the absence of the organizational capability the document assumes.

This is the only academic measurement in the theme and it reframes the whole question. Carnegie Mellon SEI, with Accenture Research, surveyed 600 professionals in January 2026: 61% report a formal AI strategy that has not been completely implemented, and only 31% use business value and ROI to prioritise what goes on the roadmap. The blockers they name are not appetite or funding but data (35%), people who know how (34%) and legacy integration (32%). The argument SEI draws from that is the useful part: the question is not which AI to adopt but which organizational capabilities have to exist before any of it returns value consistently. A strategy that names technologies and not capabilities cannot be implemented, which is why so many are not.

How to fix it — 2 approaches, 5 steps

Write the strategy as capabilities, not technologies

Rewrite the roadmap so each line is an organizational capability with an owner, not a tool with a launch date. A capability can be assessed as present or absent; a tool can only be bought.

Done when Every roadmap line names an organizational capability and who builds it rather than a tool and a launch date, each is scored present, partial or absent, and the two weakest are funded ahead of any new deployment.

  1. Restate each roadmap item as the capability it requires - data readiness, evaluation, escalation, retraining - and name who builds it.0-30 daysCIO
  2. Score each capability present, partial or absent, and publish the honest answer rather than the plan.30-90 daysCIO
  3. Fund the two weakest before funding another deployment.90-180 daysCFO

Prioritise the roadmap by value, and say so

If value and return are not the ordering principle, something else is - usually whichever function asked loudest. Make the ordering explicit.

Done when The prioritisation rule is stated in writing, the current roadmap has been re-ordered against it, and the record shows what moved down and why.

  1. State the prioritisation rule in writing and apply it to the current roadmap retrospectively.0-30 daysCFO
  2. Reorder, and record what moved down and why.30-90 daysCEO
The evidence — 3 documents
OrganizationDocumentPosition
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading Puts 61% of organizations with a formal strategy they have not fully implemented, and only 31% prioritising their roadmap by business value and return - and names data, skills and legacy integration as the top three things holding adoption back.The scale of the challenge; survey of 600 professionalsnames it
DeloitteConsultancy · March 2026Deloitte's AgenticAdopt frameworkOur reading Qualifies where the capability gap sits: just under half have begun changing how they train and reskill people, which it calls a significant risk to carrying the transformation through - so the field is split rather than uniformly behind.Building a culture of continuous learningqualifies it
McKinsey & CompanyConsultancy · August 2026Agentic AI change leadershipOur reading Places the leadership task in designing for what people actually fear, which is one of the capabilities a technology-first strategy never lists.Leaders who name and design for the fearsproposes a fix

Issue 031 organization name itnewest evidence Apr 2026

Fluency is delivered as a programme rather than maintained as a capability

Training is run once at rollout against a model generation that will be replaced within a year, and nothing refreshes it.

How to fix it — 1 approach, 3 steps

Fund capability as a standing line, not a launch cost

A one-off training budget produces a workforce fluent in a model generation that no longer exists. Fund the refresh.

Done when Capability sits on an annual operating line with a named owner rather than in a project budget, its refresh cadence is tied to model or workflow change, and coverage is reported beside deployment coverage.

  1. Move capability from project cost to an annual operating line with a named owner.0-30 daysCFO
  2. Set a refresh cadence tied to model or workflow change, not to the calendar.30-90 daysCHRO
  3. Report capability coverage alongside deployment coverage.ongoingCHRO
The evidence — 3 documents
OrganizationDocumentPosition
Boston Consulting GroupConsultancy · December 2025Targets Over Tools: The Mandate for AI TransformationOur reading Treats board fluency as maintained rather than acquired - built deliberately, topped up on a rhythm and tied to outcomes, through devices like annual retreats and quarterly immersions.Upskill with intentnames it
Boston Consulting GroupConsultancy · November 2025To invest in and upskill peopleOur reading Positions capability investment as the condition for realising value rather than as a supporting activity.Investing in people to unlock AI valuenames it
KPMGConsultancy · April 2026Building the AI business caseOur reading Includes people cost alongside hardware and software in the investment assessment rather than treating it as overhead.Building the AI business caseproposes a fix

Issue 041 organization name it

Every level is treated as a rung to climb, when a lower one is sometimes the right answer

A maturity model is read as a ladder where higher is always better, so effort goes into advancing every area rather than deciding which areas justify the investment at all.

The source of the model is explicit that its levels are not mandatory stages: a unit may deliberately stay lower because further investment would return little, and may accelerate in specific areas while holding others back on purpose. That turns the model from a scorecard into a planning tool, and it is the reading most organizations do not take, because a low score feels like a failure to be corrected rather than a choice to be made. The useful discipline is to decide the target level per area, in advance, and to be able to say why a lower one is adequate.

How to fix it — 1 approach, 2 steps

Set a target level for each area and justify anything below it

Decide, per area, what level is adequate for this organization, and record why - so a low score is either a plan or a gap, never ambiguous.

Done when Each area has a written target level with one line on why it is enough, and every area below its target is marked as a gap rather than left ambiguous.

  1. Set the target level for each area, and write one line on why it is enough.0-30 daysCEO
  2. Mark every area below its target as a gap, and everything else as deliberate.30-90 daysHead of transformation
The evidence — 2 documents
OrganizationDocumentPosition
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading States the levels should not be read as mandatory stages of advancement: a unit may intentionally remain lower where further investment would deliver limited business value, or accelerate in some areas while holding others, according to its priorities.Levels are not mandatory stagesnames it
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading Names it directly: reaching the top level everywhere is called a misconception, the model is explicitly not a compliance ladder, the most effective target is not necessarily the highest, and the target is expected to differ across areas.Prioritising by business valuenames it

Issue 05Our analysis

What a maturity average can hide

Capability is scored across several areas and reported as a total or a mean. A strong average with one weak area reads as progress, when the weak area may be what sets the ceiling.

The published method is deliberately harsher than an average and says so: a dimension is rated at its lowest-ranking capability area, and that area at its lowest-ranking indicator, even when every other goal is satisfied. The reasoning is that the areas were chosen to be the ones that matter, so a gap in any of them is a real constraint rather than something the others compensate for. An organization scoring well overall with nothing in place for one area is not four-fifths of the way there; it is held at the level of that area, and averaging is what conceals it. No organization in this index reports this happening. The caution is drawn from how the scores are constructed rather than from a published finding, and the consensus count is zero for that reason.

How to fix it — 1 approach, 3 steps

Report the weakest area beside the total, and plan against it

Publish the weakest area alongside any overall score, and direct the next investment there rather than at whatever is easiest to improve.

Done when Every published score shows the weakest area beside the total, the next block of investment is directed at that area, and a re-score on the same basis is scheduled.

  1. Score each area separately and record the lowest, not only the total.0-30 daysHead of transformation
  2. Direct the next block of investment at the lowest-scoring area.30-90 daysCEO
  3. Re-score on the same basis at a fixed interval so movement is comparable.ongoingHead of transformation
The evidence — 2 documents
OrganizationDocumentPosition
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading Rates a dimension at its lowest-ranking capability area and an area at its lowest-ranking indicator even when all other goals are met, describing this as a harsh mechanism justified by the care taken in choosing the areas.Measurement framework and scoringproposes a fix
Carnegie Mellon SEIAcademicThe AI Adoption Maturity ModelOur reading Derives each rating from more than one stream - assessor evaluation, cross-validation against survey data from the unit, and verification - rather than from a single self-report.Ratings from several evidence streamsproposes a fix

Who is represented

This dossier is drawn from 27 organizations working on the subject, 6 of which are cited directly in the issues above.

Consultancy — 12

Boston Consulting Group 14 KPMG 6 Deloitte 5 McKinsey & Company 3 Accenture 2 Capgemini 5 Arthur D. Little 1 Genpact 1 PwC 1 Teneo 1 The Hackett Group 1 UST 1

Institution — 5

World Economic Forum 3 Australian Institute of Company Directors 1 Marketing AI Institute 1 NIST 1 The Conference Board 1

Academic — 2

Carnegie Mellon SEI 1 National Bureau of Economic Research 1

Hyperscaler — 6

IBM 7 Microsoft 4 AWS 2 Google Cloud 1 Lenovo 1 ServiceNow 1

Vendor — 2

Eightfold AI 2 TechWolf 1