I want to tell you about two consultants. Both work at my firm. Both are smart. Both work hard. But their careers are heading in very different directions, and the reason has nothing to do with talent.
Consultant A is a machine. She can crank out a 50-slide deck in a weekend. Her research is meticulous. Her formatting is flawless. For years, she was our best performer — the person clients specifically requested because she delivered.
Consultant B used to be average. Solid, not spectacular. His decks were fine but not beautiful. He took longer than Consultant A on most tasks.
Then AI happened.
Consultant B started experimenting with AI early — not as a novelty, but as a fundamental rethinking of his workflow. He stopped trying to produce every deliverable himself. Instead, he started orchestrating: using AI for the research phase, the first-draft phase, the data-analysis phase, and the formatting phase. He spent his own time on the parts AI couldn't do — understanding what the client actually needed (often different from what they asked for), identifying the strategic insight buried in the data, and having the difficult conversations that moved projects forward.
Within six months, Consultant B was producing work at twice the volume and noticeably higher strategic quality than Consultant A. Not because he got smarter. Because he changed how he worked.
Consultant A is still producing everything the old way. Her quality is the same as it always was — which means she's now falling behind, because "same quality, same speed" is no longer competitive when your colleague is producing "better quality, twice the speed."
This is the Doing → Orchestrating shift in action. And it's happening everywhere.
The Identity Crisis Nobody Talks About
Here's the part that business articles about AI usually skip: this shift is an identity crisis.
If you've spent 10 years getting really, really good at doing a particular kind of work — writing code, building financial models, producing marketing campaigns, writing legal memos — your competence IS your identity. It's not just what you do. It's who you are.
And now someone is telling you that the "doing" part is becoming less valuable. That you need to shift to "orchestrating" — which feels, to many people, like cheating. Like not really doing the work.
I hear this constantly in my workshops. Seasoned professionals who feel viscerally uncomfortable delegating work to AI. Not because they don't trust the quality (though that's there too) — but because producing the work themselves is core to how they see their professional worth.
One senior engineer told me: "If I'm not writing the code, what am I even doing here?"
That question haunts a lot of people right now. And it deserves a real answer.
What You're "Doing" When You're Orchestrating
Here's my honest answer to that engineer: you're doing the hard part.
Let me break down what orchestration actually involves when it's done well. It's not "asking AI to do your job for you." It's more like being a film director.
You choose what to make. A director doesn't randomly film whatever's in front of the camera. They decide what story to tell, what matters, what the audience needs. In work terms, this is the ability to define the right problem — which, in my experience, is the most undervalued skill in any organization. Most teams are solving the wrong problem extremely well.
You assemble the right capabilities. A director chooses specific actors, cinematographers, editors — each chosen for what they bring to this particular project. In work terms, this is knowing which AI tools, which data sources, which frameworks, and which human team members to bring together for a specific challenge.
You provide the context AI doesn't have. This is huge. AI knows a lot. But it doesn't know your situation. It doesn't know that the CFO is skeptical about this project. It doesn't know that the previous three attempts at this kind of initiative failed for political reasons. It doesn't know that the data from Q2 is unreliable because of a systems migration. YOU know these things. And providing them transforms generic AI output into specific, useful output.
You evaluate and refine relentlessly. AI gives you a first draft. You read it with expert eyes and say "this assumption is wrong," "this recommendation would never work here because of regulatory constraints," "this analysis is missing the most important variable." The AI adjusts. You evaluate again. This cycle — generate, evaluate, refine — is the heartbeat of orchestration.
You make the judgment calls. At every stage, there are decisions that require human judgment. Which of these three strategies is actually right for this client? Is this risk worth taking? Does this message strike the right tone? These are decisions that require understanding consequences, weighing values, and taking responsibility. AI can inform these decisions. It cannot make them.
This is not "doing less." This is doing differently. And for most people, it's actually harder than the old way — because it requires a level of strategic clarity and evaluative skill that pure execution never demanded.
The Practical Transition
Okay, so how do you actually make this shift? Here's what's worked for the professionals I've coached:
Start with your most repetitive task. Don't try to orchestrate your most complex, judgment-heavy work first. Start with the task you do every week that follows roughly the same pattern. For a consultant, that might be the research phase. For a marketer, the first-draft phase. For an analyst, the data-cleaning phase. Orchestrate that first. Get comfortable. Then expand.
Keep a "judgment log." Every time you evaluate AI output and change something, write down why you changed it. "AI assumed flat growth; actual market is contracting." "Tone was too formal for this client." "Missed the regulatory constraint on cross-border data." Over time, this log becomes your own orchestration playbook — and a vivid reminder that your judgment is what makes the output valuable.
Set a "first human hour" rule. Before touching any AI tool, spend the first hour of any new project thinking. Define the problem. Identify the key questions. List the constraints. Determine what "good" looks like. Then bring in AI. This prevents the common trap of using AI to rapidly produce work that's aimed at the wrong target.
Build AI into your workflow, not beside it. The mistake most people make is using AI as a separate step — doing their normal work, then "checking with AI," or using AI to produce something and then manually fixing it. The shift is to integrate AI into the flow itself: AI researches while you define; AI drafts while you outline the logic; AI formats while you review.
The People Who Will Struggle
I want to be honest about something. This shift will not be easy for everyone.
The people who will struggle most are those whose professional identity is built entirely around execution quality — and who resist changing that identity.
If you're a brilliant writer who defines yourself by the beauty of your prose, the idea that AI can produce a passable first draft feels threatening. If you're a meticulous analyst whose pride comes from the accuracy of your spreadsheets, AI's ability to build models in minutes is unsettling.
These feelings are valid. But they need to be worked through, not surrendered to.
The writer who embraces orchestration doesn't stop writing. She becomes a better writer — one who spends her time on the creative and strategic decisions that make writing genuinely great, rather than on the mechanical parts that always felt like grunt work anyway.
The analyst who embraces orchestration doesn't stop analyzing. He becomes a better analyst — one who builds ten models in the time it used to take to build one, and spends his energy on interpreting what the models mean instead of wrestling with formulas.
The shift isn't from doing to not doing. It's from doing everything yourself to directing a system that does more, so you can focus on what only you can do.
That's not a downgrade. That's a superpower.