Which AI Archetype Is Your Company — Starter, Scaler, or Transformer?
Last reviewed: 23 July 2026
Three archetypes summarize AI maturity: AI Starter (no production use case), AI Scaler (1–3 use cases, scaling still open), and AI Transformer (5+ use cases, strategically differentiating). No archetype is better — each has different priorities. The decisive test for level 3: does the use case still run without its core people?
What really distinguishes the three archetypes?
An archetype describes company type and AI maturity in a simplified but reliable way — it helps prioritize recommendations without re-explaining context every time. The three types differ mainly in the number of production use cases, data strategy, and governance maturity.
No production use case
Usually just one PoC in-house. Data sits in silos, no AI-specific governance exists. Main question: where to start, what to prioritize?
1–3 use cases live
Scaling is still pending. Data strategy under construction, governance rudimentary and project-based. Main question: how do I scale what I've achieved?
5+ use cases, enterprise-wide
AI runs in critical processes. Governed data fabric in place, enterprise AI governance established. Main question: how do I differentiate strategically?
How do I precisely assess maturity — beyond the three archetypes?
Maturity models often fail when level 3 becomes a catch-all for anything mediocre. The following 5-level model defines each stage with measurable criteria instead of gut feeling — from zero production use cases to strategic differentiation that shows up measurably in EBIT.
No active AI project, AI is watched but not prioritized. No budget, no owner, no data strategy. Measurable: 0 production use cases, 0 AI roles.
PoC running or completed, no production use. Pilots fail to scale due to governance gaps — "Pilot Purgatory". Measurable: 0–1 use case, data readiness below 30%.
At least 1 use case live, but limited scaling. Sharpening test: does it run without its core people? No means level 2. Measurable: RACI in place, data quality KPIs defined.
3 or more use cases live, MLOps established, governance operating in practice. AI is reproducible, no longer dependent on individuals. Measurable: ROI measurable.
AI is no longer a project but the operating model. Agents orchestrate end to end, data is trustworthy. Measurable: AI as a KPI driver, EBIT impact.
Maturity models fail when level 3 becomes a catch-all for anything mediocre — the sharpening test is simple: does the use case run without its core people? If not, it's level 2, not level 3. Equally dangerous is the average score: a company with high infrastructure maturity and low governance maturity has a fundamentally different problem than one at medium level across all dimensions. Always act on the weakest dimension — not the average.
Glossary
Where does your company really stand? We show you in 10 minutes.
The maturity assessment maps you to an archetype and shows which dimension needs attention first — before you invest in scaling.
Start the AI Readiness Assessment →FAQ
Is an AI archetype a value judgment — is Transformer automatically better?
No. Each archetype has different priorities, quick wins, and risks. A Starter trying to act like a Transformer typically fails due to a missing foundation.
How do I tell if a use case has really reached level 3?
By the sharpening test: does it keep running without its core people? If not, it's still level 2 — a pilot stuck in Pilot Purgatory.
What is the most common mistake in self-assessing maturity?
Computing an average score across all dimensions. High infrastructure maturity with low governance maturity is a different problem than medium everywhere — act on the weakest dimension.
Do I need to develop all six dimensions equally to scale?
No, but the weakest dimension caps your overall pace. Scaling to level 4 requires that no dimension lags critically behind.
Daniel Ostner
From Chapters 1 and 2 of the Enterprise AI Guide book
Related guides
Why AI Projects Fail
88% of companies use AI, only 5.5% see measurable ROI. The three real causes: missing data foundations, no workflow redesign, missing governance.
AI Readiness Quick Scan
AI readiness has six independent dimensions. Data quality alone is behind 85% of abandoned projects. The 10-minute quick scan shows where you stand.