This is the last lesson of the program, and it is the one where everything you have built becomes a single executable act. For 110 lessons you have accumulated capabilities: spotting a broken mock at L1, running an AI-augmented sprint at L2, shipping a brand-system-as-code at L3, writing a vendor decision memo at L4, taking a seat in AI governance and naming what stays hard at L5. None of that matters if it stays in your head as competence rather than landing in your organization as change. The capstone is the plan that lands it: a real, dated, executable 90-day transformation plan for your design org's AI posture, structured as thirty days of audit, thirty of pilot, thirty of scale, that pulls together every thread of the program into one document you could hand to your CPO on Monday and start executing. The artifact is a published plan with a kickoff date, and the date is the point: a transformation plan without a kickoff date is a strategy document, and strategy documents do not transform anything. This one starts on a real Monday.

Why Ninety Days, and Why Audit-Pilot-Scale

Ninety days is chosen deliberately, because it is long enough to produce real change and short enough to force focus and survive the attention span of an executive sponsor. A twelve-month transformation plan is a plan that drifts; the sponsor's attention wanders, the priorities shift, and the plan becomes a document nobody is accountable to. Ninety days is a quarter, the unit organizations actually plan and measure in, which means your plan maps onto the rhythm the business already runs on and can be reported against at a quarterly review. The constraint is a feature: a 90-day plan forces you to pick the few moves that matter rather than listing everything you could do.

The audit-pilot-scale shape is not arbitrary either; it is the only sequence that respects how organizational change actually works. You cannot pilot intelligently until you know where you are, so the first thirty days are audit. You cannot scale safely until you have a pilot that worked, so the middle thirty are pilot. And scaling is its own distinct discipline that you do only after the pilot has earned it, so the last thirty are scale. Teams that skip the audit pilot the wrong thing; teams that skip the pilot scale a failure; teams that never reach scale produce a successful experiment that changes nothing. The sequence is a guard against all three failures, and it is the same audit-then-act discipline the whole program has taught, now applied to the org itself.

Days 1 to 30: The Audit

The first thirty days establish the truth of where your design org actually is, because a transformation plan built on an assumed starting point transforms the wrong thing. This is the L4 maturity-assessment work given a deadline. The audit covers the dimensions the program has taught you to read: the team's actual AI fluency (not what they claim, what they demonstrate), the tool stack and its lock-in and cost reality, the state of the design system and whether it is machine-readable, the governance and compliance posture, the brand-distinctiveness and provenance situation, and the craft fundamentals you are protecting.

The audit's output is not a report nobody reads; it is a prioritized, honest map of the gaps, the same heat-map the L4 maturity assessment produced, now with the specific gaps that the pilot will address identified. The discipline of the audit is honesty, because the temptation is to audit toward the conclusion you already want to pilot, and an audit that confirms your prior is worthless. You are looking for the real highest-leverage gap, the one where a 30-day pilot could produce a visible, defensible win that justifies the scale phase. The audit ends with a decision: which single gap the pilot will address, chosen because it is high-impact, pilotable in thirty days, and likely to produce the evidence that earns the scale phase.

The Audit's Concrete Deliverables

By day thirty you should have three things in hand. A maturity heat-map showing where the org is strong and weak across the dimensions. A prioritized gap list with the highest-leverage gap identified and justified. And a pilot decision: a specific, scoped intervention the next thirty days will run, with a stated success metric defined now, before the pilot starts, so you cannot move the goalposts later. Defining the success metric during the audit rather than after the pilot is the discipline that keeps the pilot honest, and it connects directly to the measurement-framework requirement the capstone demands.

Days 31 to 60: The Pilot

The middle thirty days run the pilot the audit chose, and the pilot is where the program's whole toolkit gets applied to a real, scoped intervention. The pilot has to be small enough to complete in thirty days and real enough to produce evidence, which is a genuine design constraint. A good pilot is one team, one workflow, one tool decision, or one governance contribution, run end to end with the measurement defined in the audit phase, not a sprawling initiative that cannot finish.

The capstone requires the pilot to include specific elements drawn from across the program, because a transformation plan that does not touch the program's core moves is not a design-AI transformation. It needs a tooling decision (the L4 vendor work: which tool, chosen and defended). It needs a hiring or training move (the L4 and L5 people work: a Design Engineer pilot, a craft-protecting skill program, a JD). It needs a governance contribution (the L5.3 work: a charter memo, a compliance one-pager, a provenance policy). These are not optional add-ons; they are the threads the capstone exists to pull together, and a pilot that includes a tooling decision, a people move, and a governance contribution is a pilot that exercises the program's full range rather than a single narrow capability.

The Pilot's Discipline: Kill Criteria and Honest Measurement

The pilot must carry the discipline the L5.2 pilot-portfolio work taught: explicit kill criteria defined before it starts, so a pilot that is failing gets killed honestly rather than escalated out of sunk-cost loyalty. The measurement defined in the audit phase is run against the pilot's real results, and the results are reported honestly, including if the pilot did not work, because a transformation leader who only reports successes is not trusted and learns nothing. The pilot phase ends with an honest verdict: did this work, by the metric defined before it started, and is the evidence strong enough to justify scaling. That verdict, not a foregone conclusion, gates the scale phase.

A transformation plan without a kickoff date is a strategy document, and strategy documents do not transform anything. The date is the point. This plan starts on a real Monday, and the discipline of audit-then-pilot-then-scale is the same verify-before-you-act discipline the whole program has taught, now applied to your organization instead of a mock.

Days 61 to 90: The Scale

The last thirty days scale the pilot that earned it, and scaling is the discipline the L5.2 scaling-a-pilot work taught: a successful pilot does not scale itself, and scaling badly breaks both the new initiative and the original team. The scale phase takes the validated intervention from one team or one workflow to the function, with the four things scaling requires: the tooling extended with training so the second team is not abandoned to figure it out, the training or hiring move generalized beyond the pilot team, the governance contribution moved from a draft to an adopted standard, and the measurement extended so you can tell at the function level whether the scale is working as the pilot did.

The scale phase is also where the external-voice commitment and the portfolio narrative land, because the capstone requires both. The transformation produces material: the pilot result, the governance contribution, the scaling story are exactly the genuine work the L5.6 public-voice plan routes to the four venues and the AI-honest portfolio case study documents. Scaling is not only an internal act; it is the moment the transformation becomes something you can speak and write about credibly, which is why the capstone bundles the external-voice commitment into the scale phase rather than treating it as separate. The 90-day plan ends with a transformation that has produced visible internal change and the external-facing material that turns that change into credibility.

Protecting the Original Team During Scale

The most underestimated risk in the scale phase is that the act of scaling damages the team whose pilot earned it. The mechanics are quiet and predictable: the pilot team's tooling gets churned to accommodate the function, their best people get pulled to train everyone else, their workflow gets destabilized to make it generalizable, and the source of the success is hollowed out in the name of spreading it. A scale that breaks its own proof-of-concept has defeated itself, and this is precisely the failure the L5.2 scaling work warned against. So the capstone treats original-team health as an explicit scale-phase deliverable and a thing you measure, not a hope. The function-level rollout must add capacity to other teams without extracting it punitively from the pilot team, the generalization must preserve what made the pilot work rather than diluting it into a lowest-common-denominator version, and a check on the pilot team's continued health is part of the verdict on whether the scale succeeded. The discipline is to treat the original team as an asset to protect, not a resource to consume, because a transformation that cannibalizes its own foundation is not a transformation; it is a regression wearing the language of growth.

The Six Required Components the Capstone Must Contain

The capstone specification is explicit about what a complete plan contains, and naming the six components is how you check your plan is whole rather than partial. The plan must include a tooling decision (which tool, chosen and defended, from the L4 vendor work). A hiring or training move (a people decision, from the L4 and L5 work). A governance contribution (a charter memo, compliance one-pager, or provenance policy, from L5.3). An external-voice commitment (a named submission from the L5.6 public-voice plan). A measurement framework (the metrics, defined up front, that tell you whether each phase worked). And a rewritten portfolio case study (the AI-honest narrative from L5.6, documenting the transformation itself).

Notice what this list is: it is the entire L4 and L5 curriculum, expressed as the components of one executable plan. The tooling decision is L4 vendor selection. The people move is L4 hiring and L5 roles. The governance contribution is L5.3. The external voice and portfolio are L5.6. The measurement framework is the verify-everything discipline running underneath all of it. The capstone is not a new skill; it is the integration of every skill the program built into a single act, which is exactly what a capstone should be. A plan missing any of the six is missing a thread the program spent levels building, and the check against the six is the check that your transformation is whole.

Defending the Plan, and Why the Live Presentation Matters

The capstone is defended in a live cohort presentation, and the defense is not a formality; it is the final test that the plan is real rather than aspirational. A plan that survives a hostile, informed read, peers who will ask "what is your kill criterion for the pilot," "how do you know the audit gap is the real one," "what happens at the function level if the scale breaks the original team", is a plan that can survive the actual executive room it is built for. The cohort presentation is a rehearsal for the conversation with your CPO, and the questions your peers ask are the questions your CPO will ask, which is why the defense matters.

The defense also enforces the honesty the whole program has demanded. A plan with a vague success metric, a pilot with no kill criterion, an audit that conveniently confirmed the pilot you wanted, a scale phase that ignores the original team, these are the weaknesses a live defense exposes that a written plan can hide. The act of defending the plan to people who will push back is the same forcing function the published essay and the portfolio case study used: a claim that survives a skeptic is a claim you can stand behind, and a transformation plan you can defend is a transformation plan you can execute. The presentation is where the plan stops being a document and becomes a commitment.

The Failure Modes of a 90-Day Transformation Plan

Three failure modes recur, and they map onto the three phases. The first is the undated plan: the beautiful strategy document with no kickoff date, which never starts, because a plan without a date is an intention and intentions do not transform organizations. The fix is the kickoff date, real and committed, which is why the artifact specification demands it. The second is the audit-skipping plan: the plan that jumps straight to piloting the leader's pet initiative without honestly establishing where the org is, which pilots the wrong thing and scales it into the wrong transformation. The fix is the disciplined, honest audit with the success metric defined before the pilot. The third is the never-scales plan: the plan that runs a successful pilot and stops, producing a nice experiment that changes nothing at the function level, which is the most common and most disappointing failure because it feels like success. The fix is treating scale as a distinct, planned discipline with its own thirty days and its own deliverables, not as something that happens automatically after a good pilot.

The throughline, and the throughline of the entire program, is that capability is not change. Everything you learned across 110 lessons is latent until it lands in an organization, and the 90-day plan is the mechanism that lands it. The program began with a designer at midnight wondering whether their craft was still load-bearing; it ends with that designer holding a dated, defensible, executable plan to transform how their organization uses AI, built entirely from the capabilities the program gave them. The plan is the proof that the answer to the midnight question is yes, the craft is load-bearing, and here, precisely, is how it bears the load. Set the kickoff date. The transformation starts on a Monday you can name.

Key Takeaways

• The capstone turns 110 lessons of accumulated capability into a single executable act: a real, dated, 90-day transformation plan for your design org's AI posture, structured as thirty days of audit, thirty of pilot, thirty of scale. The kickoff date is the point - a plan without a date is a strategy document, and strategy documents transform nothing.
• Ninety days is chosen because it is long enough for real change, short enough to force focus and survive an executive sponsor's attention, and maps onto the quarter the business already plans in. Audit-pilot-scale is the only sequence that respects how change works: skip the audit and you pilot the wrong thing, skip the pilot and you scale a failure, never reach scale and you produce an experiment that changes nothing.
• Days 1 to 30, the audit (L4 maturity assessment with a deadline): honestly map the org across fluency, tool stack, design system, governance, brand and provenance, and craft. Deliverables: a maturity heat-map, a prioritized gap list with the highest-leverage gap justified, and a pilot decision with its success metric defined now, before the pilot, so the goalposts cannot move.
• Days 31 to 60, the pilot (the program's toolkit on one scoped intervention): small enough to finish, real enough to produce evidence, and required to include a tooling decision, a hiring or training move, and a governance contribution - the threads the capstone exists to pull together. Carry explicit kill criteria defined before it starts, measure against the audit's metric, and end with an honest verdict that gates the scale phase.
• Days 61 to 90, the scale (the L5.2 scaling discipline): take the validated intervention to the function with training so the second team is not abandoned, the people move generalized, the governance contribution moved from draft to adopted standard, and function-level measurement. This is where the external-voice commitment and the portfolio case study land, because the transformation produces the genuine material the L5.6 public voice routes and the portfolio documents.
• The six required components are the entire L4-L5 curriculum as one plan: a tooling decision, a hiring or training move, a governance contribution, an external-voice commitment, a measurement framework, and a rewritten portfolio case study. Defend it in a live presentation, because the peers' questions are the CPO's questions and a plan that survives a hostile read is one you can execute. Capability is not change; the dated plan is the mechanism that lands the program's capability in an organization. Set the kickoff date - the transformation starts on a Monday you can name.