The AI Maturity Ladder: Four Rungs

Why Is the Question Being Asked Now?

Four developments sharpened the maturity question.

Tool abundance broke the sense of sequence

A new tool every week, and every tool promising transformation by Friday. Abundance breeds the illusion that the ladder is an elevator: subscribe, then produce. But a tool is not the rung; it is the handrail — it eases the grip, it does not replace the climb. The more tools multiplied, the more sequencing knowledge appreciated.

A state published its own ladder

The plan phases the country’s climb: awareness and data order first, infrastructure and adoption next, production and export after, governance throughout. When a state commanding billions declares “we too will climb in order,” it holds a mirror up to every business: if rung-skipping were a shortcut, the biggest player would have taken it. It is not a shortcut; it is a cliff, and now that is official.

Failed first attempts are making second chances harder

The invoice of the early-enthusiasm era is being cashed: unprepared projects have gone to the shelf, and “we tried it, it didn’t work” has settled into management rooms. The problem was rarely the technology; it was the sequence — most projects that never crossed from demo to production jumped from the awareness rung straight to the production rung. A second attempt now carries the first one’s resentment on its back.

Competition is opening up between rungs

Picture two businesses in the same sector: one has trained its team, ordered its data and automated three small processes; the other still says “we’ll look into it.” The gap looks small today — one rung. But ladder distances compound: every step on the higher rung opens doors the lower one cannot see. Rung difference accrues like compound interest.

What Is Wrong?

Four maxims muddy the ladder conversation.

“We’ve adopted AI” — the single-move fallacy

The most common sentence, the most wrong model: there is no threshold to cross, only rungs to climb. Buying a subscription is not “adopting”; it is the first stone of the awareness rung. The single-move thinker hits the first flat wall and declares “we adopted it but nothing changed” — nothing changed because the ladder stopped being climbed.

“Grand vision first, steps later”

It sounds strategic; in practice it postpones the climb. Six months pass writing the vision document, the tools change, the document goes stale. The ladder’s logic runs the other way: the small concrete rung precedes the large abstract vision. Vision clarifies by itself from the third rung — from the ground floor, all you see is fog.

“We’re not a tech firm; the Produce rung isn’t ours”

The error of equating Produce with writing software. At business scale, production is not training models; it is producing value with AI: a new service for the customer, new speed in a process, a new layer on the product. A restaurant’s menu optimisation is production; an accountant’s automated reconciliation service is production. The Produce rung belongs to everyone; only the product differs.

“Governance is for corporations and states”

The word conjures boardrooms; the business-floor version is everyday questions: which data goes into which tool, who approves what, who pulls the plug when something breaks? A three-person team that has built a human-approval flow is standing on the Govern rung. Leaving the rung “for later” is an invitation to the most expensive mistakes — and mistakes do not check company size.

The Real Mechanism

The four rungs at business scale, each with its one-sentence test.

Rung 1 — Be aware: a knowing team, ordered data

The business version is two jobs: the team getting to know the tools (not using yet — knowing: what they can do, cannot do, where they fail) and a data inventory (what we hold, where, in whose name). The test: can a random team member state, in two sentences, a strength and a weakness of an AI tool relevant to their work? If not, you are here — and here is not a place of shame but the place to start.

Rung 2 — Utilise: measured gains in live work

Tools enter real work: correspondence, reporting, analysis, design — speed and quality gains in existing processes. The defining word is “measured”: gains are not felt, they are counted. The test: is there at least one process where you can say “it took this long before, this long now” with a number? Unmeasured use is not use; it is habit — and like any number never wired to a decision, an unmeasured gain cannot be managed.

Rung 3 — Produce: value the customer notices

The gain spills from inside to outside: AI stops merely cutting cost and starts changing your offer — faster delivery, a new service layer, personalised product. The test: does your customer notice something that changed because of AI, even without naming it? Efficiency that stays indoors is rung two; difference that reaches outdoors is rung three.

Rung 4 — Govern: written limits, distributed authority

As use grows, the questions grow and the answers go on paper: data boundaries, approval thresholds, stop authority, the delegation ruler. The test: can a new hire learn what may and may not go to AI without asking anyone — from something written? If yes, you are back at the ladder’s base — because Govern is not the last rung but the safety rope of the next lap.

Who Is Affected, and How?

The same ladder produces four different climbs.

The micro business and the solo professional

The advantage is agility: four rungs can fit four quarters. The risk is skipping rung one as “not for me” — in a one-person business, team training is the owner’s own learning hour, and it is the most neglected investment there is. Two hours a week of learning is the micro scale’s ladder subscription.

The growing SME

The most critical profile: workload pressures it into rung-skipping — “we can’t keep up, automate it now.” Resist. The cure for not keeping up is never bolting a tool onto a disordered process; automate disorder and the disorder scales. The SME’s sequence is the plan’s sequence: data and capability first, automation next, product after.

The corporate and the family firm

The trap here is rung inflation: every department builds its own ladder, and where the company stands, nobody knows. The remedy is one map — a rung diagnosis per unit, one climb plan for the whole. And the executive floor’s own Awareness rung is not skippable: budget assigned to a technology the decision-maker never learned is a blind man buying a compass.

The agency and the consultant

The ladder is also the service provider’s sales map: sell solutions without diagnosing the client’s rung and you sell fourth-rung products to second-rung clients — and burn the relationship. The right question is never “what do you want” but “where are you” — in the language of the new contract: whoever matures the client wins, and whoever diagnoses the rung matures the client.

Decision Order

The climb plans in four steps; the sequence does not bend.

First, locate the rung

Run the four tests in order — each sits at the end of its rung above. The first “no” you hit is your rung. The locating is done honestly and without mercy: the rung-one business that believes it is on rung three holds the most expensive delusion in the building.

Then set a one-rung target

The planning horizon is one rung up — not two. “We will complete rung one in two quarters” is a measurable target; “we will become an AI company” is a wish. The rung counts as complete when its test flips to “yes” — no other definition is accepted.

Third, match the resource to the rung

Each rung eats a different resource: learning hours and data order on the first, tool subscriptions and measurement on the second, development budget on the third, writing and audit effort on the fourth. Committing resources ahead of the rung — stacking enterprise licences on rung one — is waste in its politest form.

Last, align with the state’s calendar

With rung and target clear, look at the public calendar: which programme cheapens which rung? Literacy programmes are rung one’s tailwind, GPU credits belong to rungs two and three, regulatory guides to rung four. Alignment comes last — because wind only helps a ship that has a course.

Where to Start?

The first month: four finishable jobs.

Run the rung test with the team

Four tests, one meeting: everyone answers for themselves, the result gets discussed together. Testing as a team produces more than a diagnosis — the ladder vocabulary becomes shared, and “where are we” becomes a legitimate question.

Write the one-rung target down

One sentence, one date, one owner: “By the end of two quarters, everyone on the team uses one tool in their own work, with numbers; owner: X.” An unwritten target lives on the wish shelf.

Install the weekly ladder hour

One block in the calendar: the team’s learning, trying, sharing hour. The hour is constant whatever the rung; only its content changes with the rung. A climb is just regular training under another name.

Plant the next rung’s seed today

Rungs are sequential but their preparations overlap: data order germinates on rung one, measurement infrastructure on rung two, written rules on rung three. Sowing the upper rung’s seed while standing on this one — that is the ladder’s compound interest.

What Not to Do?

The climb’s four classic falls.

Skipping a rung

The cardinal sin, once more: production without awareness, automation without measurement, scale without rules. A skipped rung does not vanish; it posts to the debt ledger and returns with interest — usually during a crisis, via the most expensive collection method available.

Camping on a rung

The opposite pole: perfectionism. The business that says “let’s perfect all the data first” turns rung one into a residence. A rung’s measure is not flawlessness but its test turning “yes” — once yes arrives, you climb; polish can be applied from the rung above.

Marching everyone in lockstep

Speed differences in a team are natural: some sprint, some grip. Setting the pace to the slowest halts the climb; setting it to the fastest drops the team. The working formation is vanguard-and-rearguard: the quick open the route and turn back to pull — a learning team is a team that pulls its own.

Letting the tool vendor draw the ladder

The vendor’s ladder is a product catalogue: every rung, a licence upgrade. Rung diagnosis and climb planning happen inside; the vendor and the tested partner add value in the plan’s craftsmanship — never its architecture.

A Solid Digital Foundation

The ladder stands on four stones.

The rung map

Four tests, four answers, one date — the business’s maturity photograph. Retaken quarterly; the difference between two photographs is your climbing speed.

The one-rung target page

Sentence, date, owner, completion condition. Its brevity is its strength: everyone knows it by heart.

The ladder hour

The weekly calendar block — learning’s institutional guarantee. Every cancelled ladder hour is a quiet rung debt.

The public tailwind list

Beside each rung, the public programme that cheapens it: literacy, credit, voucher, guide. The list feeds off the monitoring routine and informs resource decisions — whoever measures the wind trims the sail right.

Frequently Asked Questions

Is the state’s ladder really the same as ours? The state seems to have started Govern already.

Well spotted: the plan runs its four axes concurrently, not sequentially — because a state, at its scale, climbs the ladder while building it. The lesson for a business: rungs are ordered but not walled; every rung carries the seed of the one above and the maintenance of the one below. The order is the order of the centre of gravity: where does most of your effort flow today? If that question has one answer, you are on a ladder; if it splits four ways, you are still on the ground.