Accenture and Deloitte Are Selling the Same Brain to Every Company in Your Category

In October, Anthropic and Deloitte announced that Claude would be made available to more than 470,000 Deloitte people across the firm's global network, alongside a Claude Center of Excellence and a program to certify 15,000 professionals. CNBC reported the rollout spanned 150 countries and would include custom Claude personas built for different groups, from accountants to software developers.

Six weeks later, Accenture and OpenAI announced a collaboration equipping tens of thousands of Accenture professionals with ChatGPT Enterprise, naming OpenAI as one of Accenture's primary AI partners for its next generation of services. Eight days after that, Accenture signed with Anthropic as well. TechCrunch reported a three-year term, the formation of an Accenture Anthropic Business Group, and formal Claude training for 30,000 employees.

By April, Accenture was rolling Microsoft 365 Copilot out to roughly 743,000 people, which Microsoft described as the largest enterprise Copilot deployment to date. In May, PwC announced it would certify 30,000 US professionals on Claude, having already run a 200,000-seat ChatGPT Enterprise rollout. Industry tracking puts total AI investment across the Big Four and the top strategy houses at more than $10 billion since 2023.

Each of those announcements was sold as an edge over the firm next door. Stacked up, they mostly show a handful of large organizations arriving at the same shortlist of vendors within about seven months of each other.

The teardown

The most useful thing published about any of this did the obvious and apparently unpopular thing. It stripped the naming conventions off the major firms' AI frameworks and compared what was left.

The comparison found that most frameworks are packaging for the same four workstreams: assess AI maturity and risk, prioritize a use-case portfolio by ROI and feasibility, deploy on a governed platform layered over a hyperscaler, and upskill the workforce while standing up a center of excellence. It also noted that comparing the headline investment figures directly is misleading, since the numbers cover different scopes and time horizons, and that genuinely novel components are thin on the ground.

Four workstreams. Billions of dollars. A dozen brand names on top.

None of that is scandalous on its own. Convergence on best practice is what a maturing category looks like, and standardized AI governance is arguably what most enterprises should want. Similar frameworks only become a brand issue once you look at what firms are using them to produce.

Differentiation is a function of input variance

Here is the mechanism, stated plainly.

A brand's positioning is the output of a process. Someone looks at a market, decides what is true about it, decides what the company should therefore claim, and writes it down. For twenty years the variance in that output came from variance in the inputs. Different firms held different data, different analysts had different instincts, different partners had watched different things go wrong.

Now compress the inputs. The same four workstreams, executed on the same three or four frontier models, by people certified through the same vendor programs, inside firms whose internal tooling is being deliberately standardized. eMarketer flagged the consequence in a single line, noting that as big consultancies adopt the same agentic playbook, the risk of AI-driven sameness grows.

That line deserved far more attention than it got. If the category leader, the closest competitor and the challenger below them all buy transformation from firms running the same process on the same models, the market analysis converges first. Then the segmentation. Then the category framing, the positioning, the messaging. Sameness of this kind arrives without anyone choosing it, as a property of a pipeline that has stopped generating variance at the top.

And the firms selling differentiation are the mechanism producing the convergence, which is a structural outcome rather than a hypocrisy. Consulting economics have always rewarded a repeatable methodology deployed at scale, and AI made the methodology enormously more repeatable. The incentive was pointed here the whole time.

What consulting actually sold

Any competent operator could always read the framework, so the traditional defense of the Big Four rested somewhere else: on judgment. A partner who had watched forty of these programs and could tell you which one you were about to become.

Computing reported in June that Accenture has tied career progression to AI use. Reuters reported that the firm launched an $865 million restructuring in September to realign its workforce, cut costs and improve efficiency. Accenture confirmed thousands of job cuts on its earnings call, and its global headcount fell from roughly 791,000 to 779,000 across a single quarter.

Judgment of the kind that justified the fees gets accumulated by people, over years, on engagements. Where the next cohort acquires it is an open question once a model does the analytical middle of the work and the pyramid beneath the partner thins every quarter. The firms will have to solve that for themselves, and their timeline for solving it becomes yours if judgment was what you were buying.

You are buying this on purpose

There is a version of this argument that lets marketers off the hook, and it should be resisted.

Most of this convergence gets chosen by the buyer. A CMO who commissions a category redefinition from a firm running a standardized playbook, on a model that has read the same public material as every competitor's model, has picked an input optimized for defensibility. Which was always the appeal of the framework. Nobody gets fired for it.

Which leaves the practical question. If the analytical layer is converging and available to everyone, whatever the model cannot reach becomes the scarce input. Proprietary customer contact. The argument nobody else is willing to make in public. The operator with an idiosyncratic read of the market who has not yet been trained out of it. All of that used to be a nice-to-have in an era when analysis itself was expensive, and it now carries most of the remaining variance.

The firms will keep selling differentiation, and they will keep delivering the same four workstreams, and the deliverables will keep being good. Good and identical works fine as the outcome of a governance program. As the outcome of a brand strategy it is a slow, expensive way to disappear into your category.