You Paid for an AI Strategy. It's Sitting on a Drive.

I know this pattern because I see it regularly. A company paid a six-figure sum for an AI strategy. The document is 80 pages long. It's professional: benchmarks, maturity matrices, roadmaps, use cases with prioritization. The last slide says "recommended next steps." The consultants left. The document went to a drive. And there it sits.

Three months later, nobody can say what specifically from that strategy has been implemented. Because nothing has been implemented.

Why This Happens

This isn't one company's problem. It's a systemic issue with the model that large consulting firms operate under. The model works like this: a team of consultants comes in for 6-12 weeks, conducts interviews, does analysis, writes a document, presents the findings, and leaves. The deliverable is a document. Accountability ends the moment that document is handed over.

The company is left with a report telling it what it should do. But not how to do it in the context of its specific organizational culture, its specific people, its specific constraints. Because the consultants don't know that — they were there too briefly and talked to too narrow a group of people.

80 Pages, Zero Implementation

I've seen these documents. They're well-written. They have nice charts. They contain genuine insights. The problem isn't the quality of analysis. The problem is what happens after the analysis — or rather, what doesn't happen.

A typical AI strategy from a large consulting firm contains:

An organizational maturity assessment (2x2 matrix, as they love).
A list of 15-20 potential use cases scored by impact and feasibility.
An 18-24 month roadmap with phases.
Recommendations on organizational structure and capabilities.
An estimated budget (usually understated so as not to alarm anyone).

Each of these elements is valuable. None of them is sufficient. Because between "recommendation" and "implementation" lies a chasm that no document alone can bridge.

Who Implements the Strategy

And here we get to the heart of the problem. After the consultants leave, implementation falls on people inside the organization. These people have their day jobs. They don't have AI expertise. They don't have the mandate to change cross-functional processes. They don't have a budget for experimentation. And they have nobody to help them get from slide "recommended use case #7" to a working solution.

Some companies then hire a second firm — to "implement the strategy." This is a technology company that receives the 80-page document and tries to extract a project scope from it. Translating from consulting-speak into implementation language takes weeks and generates additional costs. Some recommendations turn out to be infeasible given the client's actual infrastructure. The document's roadmap doesn't survive contact with reality.

Strategy Is Not a Document

An AI strategy is not a product you can deliver and leave behind. It is a process that must be co-created with the organization and shepherded for months — from the idea, through the first pilot, to a working solution and measured results.

This is the fundamental difference between the "we deliver a document" model and the model I believe in and practice: I co-create the strategy with the client and stay to make sure it materializes.

What does that mean in practice?

The strategy is not 80 pages. It's 15 — because it focuses on what we're doing in the first three months, not on what we could do over two years.
Instead of a list of 20 use cases, there's one, chosen together, with a clear success criterion.
Instead of an 18-month roadmap, there's a quarterly plan with specific milestones.
Instead of "recommended next steps," there's joint work to make those steps actually happen.

Shared Accountability, Not Recommendations

Here's a word you won't find in proposals from large consulting firms: shared accountability. A consultant who delivers a document doesn't take responsibility for outcomes. They take responsibility for the quality of the document. Those are two different things.

My approach looks different. I co-create the AI strategy with the client — because a strategy that doesn't include the voices of people who will implement it is fiction. I stay through execution — because a strategy nobody shepherds dies a natural death. I take shared accountability for results — because if I'm not willing to put my reputation on the line, it means I don't believe in what I proposed.

Is this a comfortable model? No. It demands deeper engagement, a longer relationship, and a willingness to tell the client things they don't want to hear. But it produces outcomes, not documents.

When Strategy Actually Works

An AI strategy works when it meets three conditions:

It is co-created — with the people who will implement it, not just the board that commissioned it. Operations, IT, business — everyone must see it as meaningful, because they're the ones who will have to change how they work.
It is specific — not "we should consider deploying NLP in the customer service process," but "within 8 weeks we will build and test a ticket classifier on the last 6 months of data, measuring reduction in first response time."
It has an owner who stays — someone who doesn't leave after the presentation, but is in the project every day, solving problems, adapting the plan to reality, and accountable for the outcome.

A Question for You

If your company has an AI strategy — open that document. Check how many of the recommendations have been implemented. If the answer is "few" or "none" — it's not because your team is bad. It's because the strategy wasn't designed for implementation. It was designed for delivery.

Next time someone offers you an AI strategy, ask: "What happens after the document is handed over? Who stays? Who takes responsibility for the outcomes?" The answers to those questions will tell you more than 80 pages of a report.

If you need an AI strategy that won't end up on a drive but will translate into real change in your organization, I invite you to a conversation — Leszek Giza.