5.4.1. Iterative Transformation: Envision, Experiment, Launch, Scale

💡 First Principle: In an uncertain domain, the fastest path to large results is a sequence of small, measured steps, each informing the next. Iterative transformation replaces one big bet with a loop: imagine what could be valuable, test it cheaply, launch what works into production, scale what proves value, and feed what you learn back into the next round.

The exam guide names four phases:

• Envision: identify and prioritize opportunities aligned with business objectives, with outcome metrics and stakeholders defined (Phase 3). Work backwards from the outcome.
• Experiment: run proofs of concept and pilots with clear hypotheses, baselines, and success criteria; keep them small, fast, and cheap; decide at a stage gate (3.1.3).
• Launch: take proven pilots into production with the governance, security, monitoring, support, and change management that production requires (5.4.5).
• Scale: extend the capability across more users, processes, and business units, building reusable components and sustaining value through feedback loops (5.4.2, 5.4.4).

The loop runs at two levels: within a single initiative (each release informs the next) and across the portfolio (each wave of initiatives builds capabilities the next wave reuses). Not every experiment proceeds; stopping is a valid outcome that returns learning to the envision stage.

Note the wording. The AWS Cloud Adoption Framework for AI whitepaper describes a similar cycle as envision, align, launch, scale, where "align" emphasizes organizational alignment and foundational capabilities. The AIB-C01 exam guide uses envision, experiment, launch, scale. The ideas are compatible (alignment work happens throughout), but answer exam questions with the exam guide's terms.

⚠️ Exam Trap: The exam's iterative phases are envision, experiment, launch, scale. Options that skip experimentation (envision straight to enterprise rollout), or that treat transformation as a single linear project with no feedback loop, are distractors. Also do not confuse the exam's sequence with the CAF-AI whitepaper's "align" phase if an option lists phases.

Reflection Question: A CEO wants to "deploy generative AI across all customer channels by year end." Translate that ambition into the four phases, describing what would happen and what decision would be made at the end of each.