Part 2 of 9 — Strategy
At a board meeting last year, I watched a CEO stand up and announce that his company had "53 AI initiatives in flight." He said it like it was a triumph. The board, blessedly, knew it was a problem.
This is the most common pathology I see in agentic enterprise programs. Activity gets confused with strategy. POCs get confused with portfolio. Initiative count gets confused with impact. And somewhere in the middle of all that motion, the actual question — what are we trying to change about how this enterprise runs? — never quite gets answered.
This article is about the seven capabilities that make up the Strategy focus area in the agentic maturity framework, and the journey one company took to move from "53 things" to a coherent agentic strategy that could actually be defended in front of an executive committee, a board, and a guest.
The Seven Capabilities of Agentic Strategy
Strategy in the agentic enterprise is not one thing. It's seven, and they have to fit together:
Agentic Strategy — the actual artifact: how you identify, build, deploy, and govern autonomous agents, and how much independence each one gets
Use Case Identification — the discipline of choosing what to build, scored on business value, feasibility, and strategic fit
Business Alignment & Enterprise Readiness — the breadth of executive ownership; do you have leaders with both authority and skin in the game
Brand Alignment — agents are now a guest-facing surface; their voice, tone, and judgment is your brand
Digital Labor / Workforce Readiness — the planned ratio of human-to-agent work, role redesign, and the operating reality of mixed-team work
Value Management — how you measure and realize value, beyond "we shipped something"
Governance — the operating structure for choosing, funding, prioritizing, and retiring agent work
Companies at maturity Level 1 typically have one of these — usually a list of use cases — and call it strategy. Companies at maturity Level 5 have all seven, and they're integrated.
What Level 1 Looks Like
Low-maturity strategy has a particular smell to it. You can detect it from a distance:
Use cases are gathered from "submit your AI idea" forms across the org
Selection criteria are vague, often a 2-axis grid drawn on a whiteboard six months ago
No one owns the P&L impact of the AI portfolio at the executive level
Every business unit has its own definition of "value" — Parks measures attendance lift, Studios measures box office attribution, Networks measures viewership, Skies measures load factor, and nobody can compare any of it
Brand and tone of agents is a thing nobody is thinking about until a guest screenshots an awkward response from your most beloved character chatbot
Workforce planning is a separate stream that talks to AI strategy once a quarter, badly
Governance is a meeting that happens, gets postponed, and produces a status update
If three of those describe your enterprise, you are not at strategy maturity Level 3. You are at Level 1.5 with optimism.
Aurelia's Starting Point
When I first walked into Aurelia, they had 47 active AI POCs spread across 11 business units — from a guest concierge prototype at Aurelia Kingdom to a script analysis tool inside Aurelia Pictures to a baggage routing optimizer inside Aurelia Skies. Total combined P&L impact over the prior 18 months: a claimed $42 million in savings, of which the CFO could verify roughly $11 million. The other $31 million was "productivity gains" measured by survey.
The CDO was new. She'd been hired with a mandate to make agentic transformation real, and she had about 18 months of executive runway before patience would run thin. Her first move was the one most leaders avoid: she ran a portfolio review and killed 40 of the 47.
That decision cost her real political capital. It also bought her something more valuable: focus.
The Big Rocks Aurelia Had to Move
Rock 1: Killing what was already running
The 40 POCs Aurelia retired had champions, sunk cost, and in some cases active steering committees with members of the executive team on them. Killing them required the CDO to (a) write a clear set of portfolio criteria, (b) communicate those criteria upward to the executive committee and downward to the BU leaders simultaneously, and (c) replace the killed work with something more compelling. That last part is the one most enterprises skip.
The replacement was the Big 3: three agentic outcomes the entire enterprise would invest behind for the next 18 months.
Guest Concierge — a unified concierge agent that would meet guests across digital and physical surfaces (the streaming app, the parks app, the in-room TV at hotels, the cruise stateroom interface, the gate area at airports) and handle planning, recommendations, problem resolution, and proactive personalization
Yield & Demand Optimization — an agent that worked across ticket pricing, hotel rates, cruise fares, airfare, and merchandise, balancing demand signals against capacity in near-real-time
Production Operations — an agent embedded in the studio production pipeline that handled scheduling, location logistics, talent coordination, and post-production routing
Three is not a magic number. The point is that the number had to be small enough that the executive team — with all of its film slate reviews, park capital plans, and quarterly earnings prep — could keep it in their heads.
Rock 2: Getting BU leaders to commit P&L lines
A use case is not a strategy until someone owns the P&L impact. At Aurelia, this meant moving each Big 3 outcome onto specific BU leaders' annual operating plans, with revenue or cost numbers attached, signed off in advance.
The Guest Concierge work had four owners (Parks & Resorts, Voyages, Skies, and Aurelia+) with shared accountability. Yield was owned by the Chief Commercial Officer with co-ownership from each revenue-generating BU. Production Operations sat with the Studios president.
This was harder than killing POCs. It required redrawing accountability lines, in some cases reworking compensation, and in every case demanding that BU leaders take real career risk on agentic outcomes. Two of the BU leaders refused initially. One eventually came around. One was replaced.
Rock 3: Standardizing on one value framework
Aurelia had been using a different ROI framework in every business unit — RevPAR lift in hotels, ATV lift in parks, ARPU lift in streaming, contribution margin in airline. The CDO mandated a single overlay: every agent program had to articulate Real People Impact (the human experience metric, in this case heavily weighted toward guest NPS and cast member experience), Revenue Uplift (top-line contribution), and Cost Takeout (bottom-line contribution). All three. Always.
The discipline of forcing every program to answer all three was clarifying. It killed off "interesting but unmeasurable" work and elevated programs that had been quietly delivering on RPI without ever bragging about it — including a small experiment inside Aurelia Live that was using an agent to manage understudies and crew rosters across touring productions, and was making the touring company's stage managers measurably less miserable.
Rock 4: Brand alignment — the agent is the company
This is the rock that hospitality and entertainment companies feel more sharply than most industries. When an Aurelia guest interacts with a Guest Concierge agent inside the Kingdom park, that agent is, in that moment, the brand. The voice has to be right. The tone has to be right. The judgment has to be right. A clumsy reply from an agent in a park branded around wonder and care does more brand damage than the same reply from a generic chatbot at a bank.
Aurelia stood up an Agent Brand Council that sat across Marketing, Brand, Legal, and the Office of the CDO. The council reviewed and approved every guest-facing agent persona, voice guide, and escalation tone. They published a living "Voice of the Agent" guide that every agent shipping to production had to demonstrably comply with. The guide was 22 pages and revised quarterly.
This investment was not technical. It was also not optional.
What "Leading" Looks Like
Eighteen months in, Aurelia's strategy capability looks like this:
A quarterly portfolio review where the executive committee retires, expands, or holds each agent program based on standardized metrics
The Big 3 has grown to a Big 14, with three retired and four more in pre-pilot
Every BU leader has at least one agentic outcome on their operating plan, with the Guest Concierge mega-program shared across four BUs under a single executive sponsor
A brand voice guide for agents that is enforced at the platform level — every guest-facing agent the company ships speaks in a recognizable, reviewed tone, with persona variants for different surfaces (the parks app sounds different from Aurelia+, but both are recognizably Aurelia)
A workforce-to-agent ratio is published in the company's internal annual planning document, by function and by BU
Governance meets monthly, has teeth, and the agendas get shorter every quarter because the discipline holds
This is not "leading" because the technology is impressive. It's leading because the strategy is legible. Every cast member at Aurelia can answer the question "what is our agentic strategy?" in two sentences. Most companies fail this test in six paragraphs.
The RPI at the Other End
Strategy maturity sounds abstract until you trace it to humans. At Aurelia, the Guest Concierge program lifted post-visit NPS at the flagship Kingdom park by 9 points in its first full season — driven mostly by guests who had previously been frustrated by long lines for restaurant reservations and ride access getting proactive recommendations that actually fit their party's pace and preferences. The Production Operations agent gave assistant directors and line producers an average of 11 hours back per week. Several of them used those hours to mentor junior talent who had been waiting years for the kind of attention that a frantic schedule had been preventing.
That is the RPI signal. Not "we deployed an agent." Did the work get better for the humans on the other end?
What to Do This Week
If you want to test where your strategy actually sits, try this exercise. In the next leadership meeting, ask three questions:
What are our top three agentic outcomes for this fiscal year?
Who owns the P&L impact of each one?
What is our single, shared definition of value for agent programs?
If you cannot answer all three in under five minutes, you are at strategy maturity Level 1 or 2. That's not a verdict. It's a starting point.
Next article: Operating Model. The hidden architecture of how agent work actually gets done — across a 120,000-cast-member organization that runs film slates, theme parks, cruise ships, and an airline simultaneously.
This is Part 2 of a 9-part series on agentic enterprise maturity.