Who Owns the Codified Judgment
The claim¶
The Partnership Pyramid, Generalized asks whether the professional-services staffing model survives once AI agents supply the output that used to require an army of associates. Professional services turns out to be close to the boundary case where the pyramid does not cleanly invert, because firms need a wide junior base to train tomorrow's partners.
That leaves an open question the pyramid framing cannot answer, because it is not a staffing question. Once a firm's judgment is codified well enough that an agent can apply it, that judgment becomes an asset someone can own, sell, or buy. The staffing model can survive fully intact and this question still bites. It is a property-and-rent question, not a headcount question: once judgment is codified, who owns it, and who is exposed when someone else does?
Two things get codified, not one¶
Codifying a firm's work splits into two different assets, not one, and they behave nothing alike:
- Execution Specs — machine-readable contracts for commodity work: inputs, outputs, rules, the quality bar. AI-drafted, disposable, cheap to produce, and cheap to replace once written.
- Decision Rubrics — the proprietary judgment inside the firm's frameworks: the thresholds, the override conditions, the reasoning a senior applies that the framework alone does not capture. Hand-authored, durable, and the actual asset a client is paying for.
A firm can build either one internally, or acquire a company that has already built its equivalent. That choice, buy or build, is where the rent question gets decided.
The worked example: two paths, same destination¶
Accenture's own recent moves show both paths at once, and the two differ in kind, not just in name.
The buy path. Accenture agreed to acquire Faculty, described in its own announcement as "a leading UK-based AI native services and products business." Faculty's own product, Faculty Frontier, "will join Accenture's suite of products that help organizations make better, faster decisions by connecting data, AI models and business processes into a unified decision system" (Accenture, 2026). Read against the framework above, this is a bought Decision Rubric. Accenture did not ask its own partners to write down the judgment that makes them scarce. It bought a company that had already built an equivalent judgment engine, and folded it into its own product line.
The build path, made concrete. Separately, Accenture's expanded AI Products line with Amazon Web Services (AWS) ships "pre-built, enterprise certified AI agents that autonomously collect data, build a contextual knowledge layer, and transform manual coordination into self-managed decision intelligence across planning, logistics and inventory visibility" (Accenture, 2026). This is closer to Execution Specs made real: a defined, certified, repeatable contract for commodity coordination work, productized and sold rather than delivered by hand each time.
These two moves are not equally new, and the distinction is sourced, not asserted. An accelerator library or a starter-template offering is old ground for consulting firms — a template a consultant still assembles and runs, start to finish. Accenture's certified agent library goes further: the agents themselves perceive system state and write back autonomously, and the consultant's role shrinks to deployment and oversight rather than doing the execution by hand. Accenture describes the result as a "human-in-the-lead" model, with its own engineers "embedded" alongside clients rather than running the coordination work themselves (Accenture, 2026). That is a real narrowing of what the human does, even though a consultant has not left the picture entirely.
Faculty's product is more mixed than a single label suggests. Its core, Computational Twin, "brings together an organisation's data sources, predictive models, optimisation algorithms, and business rules into a coherent, dynamic simulation of how work and decisions flow across the enterprise" — that is classic simulation-and-optimization technology, the kind consulting firms have sold for decades, now packaged as a product instead of an engagement (Faculty, 2026). Its newest release adds a real agentic layer on top of that core, not instead of it: an AI Decision Agent (ADA) that "enables natural-language interaction, tailored recommendations, and specialist support agents." Accenture bought both layers in one acquisition rather than building either internally.
Why buying changes the fight, not just the balance sheet¶
Asking a partner to hand-author a Decision Rubric asks the person whose scarcity is their compensation to write down the exact thing that makes them irreplaceable. That is a real internal disincentive, and it is a large part of why the "build" path stays a slide deck at most firms rather than a shipped asset: no one inside the partnership is rewarded for finishing it.
Buying sidesteps that disincentive at the acquisition step. The firm does not have to convince its own partners to disintermediate themselves before the deal happens; it acquires a company that already did the equivalent work somewhere else. The fight does not vanish, though. It relocates: from an internal negotiation nobody has an incentive to start, to a post-acquisition integration question that lands on the acquiring firm's own practice-area partners once they discover whose billable work the newly bought product now replaces — with a valuation, a press release, and a named practice area attached to it.
One reading treats that relocation as a real, structurally different dynamic: an internal codification fight is diffuse and has no forcing function, while an acquisition is a datable, observable event with a specific practice area's billables now visibly at stake. A competing reading treats this as ordinary post-acquisition integration politics, the same "whose role survives the deal" question every services roll-up has faced, just described in AI-native vocabulary. Both readings are defensible from the same facts. Which one is closer to true likely depends on the specific firm and deal, not on the pattern in general.
What this means for a firm leader¶
The question Partnership Pyramid, Generalized poses, whether the pyramid inverts into an arrow, may simply be the wrong question for a firm's leadership to sit with. Accenture's own moves suggest a third option neither a pyramid nor an hourglass names: keep the staffing model exactly as it is, and bolt a separate, productized revenue line onto it.
The more useful question for a firm leader is not "will my pyramid invert." It is: which of my practice areas is closest to having its judgment bought or built out from under it, by a competitor or by my own firm's product arm, and who inside my firm is exposed when that happens. Whoever owns the answer to "is this commodity or judgment" for a given practice area is, in practice, deciding whose compensation is safe and whose is not. That call is worth making deliberately, before someone else, inside or outside the firm, makes it first.
This is a different question from whether to build or buy a given capability at all, which Build vs. Buy already covers in general. What this page adds is specific to professional-services judgment: buying does not just change cost and control, the two criteria that guide already covers. It changes who is exposed to the fallout, and when that fallout becomes visible.
Sources¶
- Accenture — Accenture to Acquire Faculty to Scale AI Capabilities (newsroom), 2026. a leading UK-based AI native services and products business built on highly technical applied AI skills. Faculty Frontier, Faculty's enterprise decision intelligence product, will join Accenture's suite of products that help organizations make better, faster decisions. View source · verified 2026-07-05 · primary
- Accenture — Accenture Expands Accenture AWS Business Group with New AI Products Capability (newsroom), 2026. pre-built, enterprise certified AI agents that autonomously collect data and build a contextual knowledge layer. This 'human-in-the-lead' model has already identified hundreds of millions in savings. These engineers work directly alongside clients. View source · verified 2026-07-05 · primary
- Faculty — Faculty Launches Frontier 3: The Next Evolution in Decision Intelligence for Enterprises Worldwide (PR Newswire), 2026. CT brings together an organisation's data sources, predictive models, optimisation algorithms, and business rules into a dynamic simulation of how work and decisions flow. ADA enables natural-language interaction, tailored recommendations, and specialist support agents that help teams query, analyse, and act on complex data. View source · verified 2026-07-05 · primary