Skip to content

AWS CAF perspectives, phases and the gen AI maturity model

Six CAF perspectives, four CAF phases and four generative AI maturity levels are three different lists. What each one is, and how to keep them apart.

6 min readBy AI Guru AcademyFacts last checked

In short

The AWS Cloud Adoption Framework (AWS CAF) has six perspectives (Business, People, Governance, Platform, Security, Operations) and four transformation phases (Envision, Align, Launch, Scale). The generative AI maturity model is a separate AWS Prescriptive Guidance model with four levels (Envision, Experiment, Launch, Scale). Two of the names repeat across lists, which is the usual source of confusion.

What are the six AWS CAF perspectives?

The six perspectives are Business, People, Governance, Platform, Security and Operations. AWS describes them as the way the framework groups its capabilities, with each perspective made up of capabilities that functionally related stakeholders own or manage.

A perspective is an area of capability, not a stage in time. AWS suggests evolving the foundational capabilities as the transformation progresses rather than tackling them all at once. A sponsor asking "who owns this?" is looking at one set of capabilities, and an engineer asking "how is it deployed and monitored?" is looking at another.

One qualification matters. The perspective list above is for the AWS CAF itself. AWS also publishes a separate CAF for AI whitepaper, which this article does not cover.

What are the four AWS CAF phases?

The four CAF phases are Envision, Align, Launch and Scale. AWS calls them iterative and incremental cloud transformation phases, so you can revisit them as you learn.

  • Envision demonstrates how cloud will help accelerate business outcomes. It identifies and prioritises transformation opportunities in line with strategic objectives, and ties initiatives to stakeholders and measurable outcomes.
  • Align identifies capability gaps across the six perspectives, finds cross-organisational dependencies and surfaces stakeholder concerns. This supports readiness strategies and change management.
  • Launch delivers pilot initiatives in production and demonstrates incremental business value. Learning from pilots helps you adjust before scaling.
  • Scale expands production pilots and business value to the desired scale, and ensures the benefits are realised and sustained.

Notice that Align is the phase AWS explicitly ties to the perspectives: it focuses on identifying capability gaps across Business, People, Governance, Platform, Security and Operations.

What are the four levels of the generative AI maturity model?

The generative AI maturity model has four levels: Envision, Experiment, Launch and Scale. It is a separate AWS Prescriptive Guidance model that shows where an organisation stands with generative AI and what the next level offers.

One qualification: the overview table labels level 4 "Scaling", while the level list says Scale.

What are the four levels of the generative AI maturity model?
LevelWhat AWS says it looks like
1. EnvisionExploring concepts, building awareness, identifying potential use cases. No formal projects or resource allocation.
2. ExperimentStructured pilots and proofs of concept, small teams, foundational and governance frameworks.
3. LaunchA few applications in production, with risk, governance and responsible AI policies and operational and support teams.
4. ScaleBroad adoption across departments, many applications in production, a formalised operating model and RACI matrix, reusable components.

AWS adds two cautions. No single level is better than another, because the right one depends on goals and readiness. And an organisation should not be labelled with one level overall, because each aspect of adoption should be considered independently.

How do the three lists compare?

They answer different questions. Perspectives describe areas of responsibility, phases describe the stages of a transformation programme, and maturity levels describe how far generative AI adoption has progressed.

How do the three lists compare?
RowCAF perspectivesCAF phasesGenerative AI maturity levels
Count644
NamesBusiness, People, Governance, Platform, Security, OperationsEnvision, Align, Launch, ScaleEnvision, Experiment, Launch, Scale
DescribesAreas of capabilityStages of a transformationProgress in generative AI adoption
Question it answersWho and what needs to be ready?What are we doing in this stage?Where are we with generative AI?

The shared names are the trap. Envision, Launch and Scale appear in both four-step lists. The differences are the second step (Align in the CAF, Experiment in the maturity model) and the fact that the maturity model is about generative AI specifically.

How do I avoid mixing them up?

Check the second step and the topic. If the list has "Align" it is the CAF phases. If it has "Experiment" it is the generative AI maturity model. If it has six items with names such as People or Security, it is the perspectives.

An illustration (invented, not a real company): a retail firm has a small team testing a chatbot on internal documents, with a pilot and a basic policy. A reader asks which "Launch" this is. Using the rules above, the firm's generative AI position is described by the maturity model, where structured pilots and early governance fit the Experiment level. The CAF Launch phase would only apply if the firm were delivering a pilot into production as part of a wider transformation programme.

A short checklist for any scenario:

  1. Is it asking about an area of the organisation, such as skills, controls or operations? Think perspectives.
  2. Is it asking about the stage of a transformation programme? Think CAF phases.
  3. Is it asking about how mature generative AI use is? Think the maturity model.
  4. Does the wording include "Align" or "Experiment"? That settles which four-step list it is.

Where do ISO standards fit?

They are separate from AWS frameworks. ISO/IEC 42001, for example, is an international standard that specifies requirements for establishing, implementing, maintaining and continually improving an AI management system within an organisation. It does not define CAF perspectives, phases or maturity levels.

Key takeaways

  • The AWS CAF has six perspectives: Business, People, Governance, Platform, Security, Operations.
  • The AWS CAF has four iterative phases: Envision, Align, Launch, Scale.
  • The generative AI maturity model has four levels: Envision, Experiment, Launch, Scale.
  • Align and Experiment are the quickest way to tell the two four-step lists apart.
  • AWS says no maturity level is inherently better, and aspects of adoption can sit at different levels.

FAQ

Are the CAF phases and the generative AI maturity levels the same thing?

No. The CAF phases are Envision, Align, Launch and Scale, and describe a transformation programme. The maturity levels are Envision, Experiment, Launch and Scale, and describe progress in generative AI adoption.

Where do the six perspectives come in during the CAF phases?

In the Align phase, AWS describes identifying capability gaps across the six perspectives. The perspectives themselves are areas of capability, not stages.

Does an organisation sit at one maturity level?

AWS says many organisations show characteristics of several levels across teams and use cases. It advises considering each aspect of adoption independently.

Do I need to memorise the detail of each phase for AIB-C01?

A sensible approach is to know the names, the order, and which list each belongs to, and to be able to explain what each stage is for. Check the official exam guide for the exact scope.

Sources

Last checked 2026-11-16

AI Guru Academy is an independent training provider and is not affiliated with or endorsed by Amazon Web Services. AWS service names are trademarks of Amazon.com, Inc. or its affiliates.

Next step: take the free AIB-C01 diagnostic at /diagnostic, then see the course at /courses/aib-c01-exam-prep.

AWS CAF perspectives, phases and the gen AI maturity model | AI Guru Academy