For decades, consultants hid behind process. Now that the process can run itself, there's nowhere left to hide.
The Setup
The big firms are moving fast. Deloitte just launched an agentic transformation practice with Google Cloud. IBM's consulting arm is selling enterprise-scale agent orchestration as a product. McKinsey's 2026 State of Organizations report devotes an entire framework to what they call the "agentic organization" — one where AI agents handle continuous workflow execution while humans define policy, monitor outliers, and make judgment calls.
The pitch to clients is compelling: 20–40% reductions in operating costs, 12–14 point EBITDA improvements, faster cycle times. And the pitch is real — when agents can research, synthesize, model, and draft at machine speed, delivery timelines compress dramatically.
But here's what the vendor decks don't linger on: when an agent completes 80% of a consulting deliverable, what exactly did the consultant contribute? And more importantly — how would you even measure it?
The Insight
We tend to assume that as AI does more of the work, the human's contribution shrinks proportionally. That's the wrong mental model.
What actually happens is the opposite: human accountability doesn't dilute — it concentrates.
Think about what a consulting engagement used to look like. A junior analyst spent three weeks gathering data, building the model, formatting the slide. A manager spent two days reviewing, adjusting, positioning. A partner spent two hours framing the recommendation for the client. The output reflected all of those layers, and the accountability was distributed across all of them.
Now run the same engagement with agents. The data is gathered in minutes. The model runs overnight. A draft deck exists by morning. The junior analyst is now an orchestrator. The manager is reviewing agent outputs, not human ones. And the partner's two hours at the end are now the majority of the human intellectual contribution to the entire workstream.
This is the accountability inversion: as AI handles more execution, the human's role becomes smaller in volume but larger in proportion. Your judgment, your framing, your quality bar, your read on the client — these no longer get blended into a team's cumulative output. They are the output.
McKinsey put it cleanly in their agentic organization research: the more fluid work becomes, the more deliberate leaders need to be about accountability. Fluidity without structure doesn't free people — it obscures who owns what, and why the outcome is what it is.
What This Means Practically
For consulting teams, this demands a hard rethink of how work is structured and evaluated.
First, recalibrate where human judgment gets applied. The most dangerous pattern in AI-augmented engagements is using agents to do the hard thinking, then having humans rubber-stamp the outputs. That's not augmentation — it's abdication. The human's role should be to set the frame before the agent runs, and to interrogate the output after. The hours in between can be largely automated. The moments of genuine intellectual contribution cannot.
Second, redesign how contribution is recognized. The traditional proxy for effort — time spent, slides built, models run — breaks down completely when agents can produce in minutes what took weeks. Forward-thinking firms are already measuring something different: how much of the time saved by AI gets reinvested into innovation, into client relationships, into judgment that wasn't possible before. That reinvestment is the new signal of a high performer.
Third, be honest about what you're actually good at. In a world where an agent can generate a competent first draft of almost anything, "being good at drafting" is no longer a differentiator. What remains irreplaceable is the ability to see what the situation actually requires — before the brief is written, before the agent is prompted, before the model runs. That upstream diagnostic capacity is where elite consultants will separate themselves. The agent executes what you frame. If your framing is weak, so is everything downstream.
Final Thoughts
If an agent could perfectly replicate your last three deliverables — the research, the analysis, the slide — what specifically would be lost?
The answer to that question is your actual professional value. It's also, increasingly, the only thing your clients are paying for.