We built Odyssey by deciding early and on purpose what we would refuse to sell. Not a deck. Not a tool. Both are the standard forms AI expertise takes in the strategy market, and both fail the person paying for them. But the reason runs deeper than either artifact. It relates to what AI is doing to the business of strategy itself.
The consulting model AI is dismantling
Two models have run strategy consulting for fifty years, and AI is dismantling both. The pyramid: partners sell, analysts produce, and leverage prints the margin. The deck: the deliverable as artifact, the slide as proof work was done. The retainer: you don’t buy an answer, you rent access to the people who have them. None of these serve the client. Each monetizes a single scarcity — the scarcity of analytical labor. That scarcity is over.
The supply side broke first. The work at the bottom of the pyramid — the slide a junior built over a weekend, the benchmark, the first-pass analysis — is now done by a model in seconds. The artifact that justified the invoice has collapsed in cost.
The more interesting break is on the demand side. The buyer can now produce the artifact himself. A CEO can generate a competent-looking deck without us, has done it, watched it fail to move the P&L, and concluded correctly that the artifact was never the point. What he lacks is not analysis but judgment about which analysis matters, and someone who will stand behind the answer when a board pushes back.
So the question for any strategy firm built after 2024 is not how to make decks faster. It is what you sell once the deck is worthless and the retainer-as-access indefensible. We answered that before we took a client.
What a Skill is
A Skill is a productized judgment.
Take a piece of strategic work that genuinely repeats — diagnosing AI readiness, prioritizing use cases, building the board narrative, measuring realized value — and treat it like any serious product. Fixed scope. Fixed price. A defined deliverable. A partner reviews it before it ships. Buyable two ways: on its own, where you know the cost and what you get before committing a euro, or embedded inside a broader advisory engagement as a productized component of a larger mandate.
That is the part the deck never gave you. The deck was a one-off; the Skill is a method. Run the readiness diagnostic this quarter and again next quarter against the same scoring, and you have a trend instead of a memory. The output is not a slide that ages. It is an instrument you keep.
But here is the discipline that makes it work, and the line we will not cross: a Skill is productized, never off-the-shelf. Everything ships inside the context of your business — your sector, your economics, your constraints — or it doesn’t ship. The product is the structure. The judgment is bespoke. We refuse to sell one without the other, because we’ve watched the entire industry fail at precisely the seam where the two are separated.
Why not agents
We could have built agents. The market wanted us to, and the engineering would have been the easy part. We didn’t, for two reasons — one about trust, one about durability.
The trust reason is the obvious one. An agent is a bet on autonomy: you hand it a goal and let it act. That bet pays when the stakes are low and the task is bounded. But our work reaches boards, investment committees, and CFOs in a questioning frame of mind, and there, autonomy is not the asset — accountability is. A use-case prioritization, a value-creation plan, a pre-LOI due diligence: these are not tasks you let a black box run unsupervised and unexplained. They are positions someone has to defend in a room full of skeptics.
The durability reason is discussed less. An agent lives inside your technology environment — wired to your stack, your data, your tools, your models, your permissions. That is its power and its fragility. Change the environment and the agent breaks: re-platform the CRM, swap a model, restructure the data, and the thing has to be rebuilt. You don’t own a capability. You own a configuration, and configurations expire.
A Skill sits one level above the plumbing. It is a method, not an integration — independent of whatever stack you happen to run this year. The judgment it encodes survives a migration, a vendor change, a model you haven’t bought yet. You can run it this quarter and again next year against the same scoring, because it was never bolted to a system that moves underneath it.
So we productized the judgment, not the autonomy, and we kept it above the stack, not inside it. Every Skill leaves the building with a human who reviewed it and stands behind it. In a market racing to remove the human from the loop and wire everything to everything, we decided that for strategic decisions the human in the loop is the loop — and the method has to outlive the machine it ran on.
Why advisory stays
This is also why we kept advisory rather than collapsing into a catalog. But not the advisory the word usually means.
The classical version arrives, learns your business, hands you a recommendation, and leaves — accountable for the quality of the document, never for what happens after it. That is exactly the advice AI makes indefensible. We kept the opposite kind. Committed advisory: we set the strategic frame, sequence which Skill runs against which constraint and in what order, and stay attached to the outcome rather than the deliverable.
A Skill answers a question well. It does not tell you which question to ask, against which constraint, in what order. That sequencing — what to do first, what to ignore, what your specific situation actually demands — is the judgment around the judgment. It is what separates a company with an AI strategy from a company with a stack of AI deliverables. And the firm that does it should be on the hook for whether it was right.
How we are paid — towards a new economic model
Strategy expertise has always been sold by time: the hour, the day, the retainer. You pay for access and effort, and you pay the same whether the work moves your P&L or dies in a drive. The incentive is to stay, not to be right. AI makes that arrangement impossible to defend: once the effort it is priced for has collapsed in cost, billing for it is billing for a scarcity that no longer exists.
So we price in three registers — and only one of them is the old model, reformed.
The Skills are productized: fixed scope, fixed price, published openly, from a one-day audit up. No meter, no hostage-to-access. You buy the piece you need, and you keep it. Transparency about price is just what conviction looks like on an invoice.
The committed advisory carries a base — a fixed engagement fee, never an hourly one. But the base does not buy our hours. It buys a partner who sets the frame, sequences the work, and stays accountable to the outcome for the life of the mandate. That is the line we insist on: the old retainer priced access to people; ours prices commitment to a result. You are not renting our time. You are buying into our attachment to whether the strategy works.
And on top of that base, wherever the value can be cleanly measured, we index part of the fee to what the work actually creates. We can contemplate this for the one reason most firms can’t: we built the Measure discipline first. A firm that cannot prove what the AI earned has no honest basis for being paid for it. Attribution is hard, and where a clean line can’t be drawn from a recommendation to an EBITDA point, we don’t pretend one runs. But where it can, our fee belongs inside the measurement, not on a meter counting our access. We are building toward that being the rule, not the exception. When you win, we win.
A PE partner will recognize the shape: a base that funds the engagement, plus a share indexed to the value created. The resemblance is deliberate — and so is the limit. We take no stake, deploy no capital, sit on no cap table. We put no money at risk, only our own fee. We are not investors; we are advisors who agreed to be paid like principals on the one thing we actually control — whether the advice works.
Why the floor isn’t the position
It would be easy to misread the entry point. The lowest-priced Skill costs less than a day of senior consulting, and a partner at a fund might glance at that and wonder whether Odyssey has gone downmarket. The opposite is true. The standalone Skill is the front door, not the floor plan. The engagements that define our work — a portfolio-wide value-creation mandate, a pre-LOI due diligence across several targets, a measurement program extended from pilots to the full estate — sit in an entirely different order of magnitude, and that is where most of our value, and most of our clients’, is created.
The accessible price is not a repositioning. It is an on-ramp. It lets a CEO or a PE partner test the judgment on a contained question before committing to the engagement that judgment usually points toward. The small thing is buyable, so the large thing can be earned rather than assumed. Our core client hasn’t changed, and neither has what we’re worth to them.
What we’re betting on
Eleven Skills sit along one framework — Frame, Decide, Create, Measure — and the order is the argument: most AI value is won upstream, long before deployment, and proven or lost at the very end, in the measurement no one built. “Faster emails” is not an answer to what the AI earned; a Value Realization Framework is. The catalog lives on the site. What matters is the spine, because a model without one is just a price list — and a spine, in the end, is only as good as the bets it rests on. Here are ours.
We are betting that the enterprise is tired of paying for brilliance it can’t reuse, and scale it can’t trust.
We are betting that for the decisions that reach a board, a named partner who stands behind the work beats an autonomous system no one can question — and a method that outlives your stack beats a configuration that breaks when the stack moves.
We are betting that judgment can be productized without being cheapened — that you can put a fixed price on serious thinking and still deliver it in full context.
We are betting that a firm willing to be paid on what the work earns will outlast a firm that bills for the time it took.
And we are betting that a CEO would rather buy exactly the piece they need, keep it, and run it again, than rent a conclusion that expires.
That is the model. Advisory sets the course. Skills do the work. A human signs. And increasingly, we are paid for the value of our work.
Strategy that operates by Monday.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.