How Much of Strategy Work Can AI Actually Replace?

The question every planner is quietly asking

If you run new business, you've had the thought: how much of the pitch strategy can I actually hand to AI before the next brief lands? The usual answer is "it's faster" — days of work done in hours. That's the easy claim. What's rare is someone measuring it, over time, on the same real strategy work. One team did — and the work they measured looks a lot like yours.

A before-and-after, two years apart

Siemens has a unit, Siemens Professional Education, that runs strategic foresight — reading where skills and business are heading so it can design training ahead of demand for the 292,000+ employees it serves. It's genuine strategy work: scan the signals, decide what they mean, call what capabilities will matter. Swap "which capabilities will matter" for "which angle will win the account," and it's the same job a planner does before a pitch.

A research team led by René Rohrbeck at EDHEC Business School observed that unit's foresight process twice, two years apart: a fully human-led cycle in 2023, and an AI-augmented one in 2025 — same organization, same kind of work, AI the only new variable. The study was published in the journal *Futures*. It's one of the few before-and-after measurements of what AI actually does to strategy work, rather than a claim about what it should do. (Rohrbeck, Szuppa & Schmidt, "Artificial Intelligence in Strategic Foresight" — preprint on SSRN)

AI didn't make everything faster

The short version of the finding: no uniform speed-up, and no automation. The overall cycle didn't get shorter at all — it's tied to the annual planning rhythm, and nobody set out to compress it. The time savings showed up *inside* tasks, not across the process.

And they clustered in specific steps: scanning the signals, consolidating the material, and drafting. Drafting a trend profile dropped from about 3.5 hours to about 1.25; routine analytical work was roughly halved. In those steps AI widened the breadth of analysis and tightened the structure.

But because AI took those steps, the process didn't get faster — the human effort got redistributed. The hours freed from gathering and drafting moved somewhere else.

Where they moved — and the limit that showed up

They moved to the part AI couldn't take: setting the frame, choosing what matters, interpreting what it means in context. The study is blunt that this didn't shift — in its own terms, authority over interpretation "did not move at all," and in the one phase that used no AI at all, practitioners rated its potential value the lowest.

That's the line, and it's a clean one. AI can gather, summarize, structure, and draft. It can't carry responsibility for what the material *means*. Or, as the study puts it: AI outputs are not outcomes. A faster, broader pile of well-organized inputs is not a strategy. (Why a polished AI draft still fails in a real pitch is the same gap from the agency side.)

The sharp twist is that widening the analysis made this *more* true, not less. More material and a broader view didn't make the strategy better on their own. More options meant the human job — choosing among them and setting the direction — mattered more, not less.

The human's job, the machine's job

This study didn't read as news to me, because it's exactly where I landed building Synapsis.

Writing an advertising pitch, the gathering, organizing, and framing of a first draft can be handed off. Deciding the strategic angle of the campaign — the one reading of the brief the pitch will stand on — cannot. That call runs on judgment and tacit knowledge, and that comes from years in the room.

So Synapsis puts the human part first. How that angle gets decided — 25+ years of that judgment, built into the engine — is set in advance. Then the model develops directions along that design. The human sets the angle; the machine runs it out. (How it turns a brief into directions.) In the study's language: the un-delegable work — setting the frame — is done up front by a person, and the delegable work — producing the draft — is left to the AI.

The way AI changes strategy work isn't "it replaces the person." It's "it redraws the line between what a person does and what a machine does." Foresight across a 292,000-person workforce or a creative strategy for a new-business pitch — the line is the same, and Synapsis works because it holds it.

For an agency, that redistribution has a number attached. The hours AI takes off gathering and drafting don't shrink the pitch calendar — they move to the part that wins or loses it: the angle. Same team, same deadline, but the judgment work that decides the pitch gets more of the week, not less. That's how a six-person shop can pitch like a bigger one.

Read more

• Why a polished AI strategy still fails in a real pitch: Why AI strategy looks great on paper but fails in the pitch
• How the engine turns a brief into three directions: Can AI write a creative strategy for a pitch?
• AI builds the deck — who builds the strategy: AI writes the deck. Who writes the creative strategy?

Frequently asked questions

How much of strategy work can AI actually replace?

Based on the Siemens study, not as much as "it's faster" implies. AI took over scanning, consolidating, and drafting — and sped those up sharply — but the overall process didn't get faster and nothing was automated. Setting the frame, choosing what matters, and interpreting it stayed with people. AI redistributed the work; it didn't replace the strategist.

Does AI make the strategy process faster overall?

Not necessarily. In the Siemens case the cycle length didn't change — the time saved inside tasks like drafting got redistributed to human judgment work, which took on more weight as AI widened the range of options to choose from.

What part of strategy work can't AI do?

Framing and interpretation — deciding what the gathered material actually means and which direction to commit to. The study's line is that "AI outputs are not outcomes": a faster, broader set of well-structured inputs is not a strategy until a person makes the call.

Can AI replace an agency planner or creative strategist?

No — but it changes the job. In the Siemens study AI took over scanning, consolidating, and drafting, while framing and interpretation stayed with people. For a planner that means AI can build the first draft of the territory; deciding which angle wins the account still runs on judgment. The headcount question isn't "fewer planners" — it's the same planners pitching more, with the gathering off their plate.

How does Synapsis divide the work between human and AI?

The un-delegable part — how you decide the strategic angle a pitch persuades from — is designed in advance by a person with 25+ years in advertising and set into the engine. The delegable part — developing directions from a brief along that design — is what the engine runs. The human sets the angle; the machine runs it out.