Pods, not pyramids

A traditional strategy engagement has a familiar shape. A senior partner sells the work and presents the answer. Below them sit managers, and below the managers, a pyramid of analysts. A traditional pod typically takes 6 to 12 weeks.1Show source for note 1 Much of that time is spent at the wide base of the pyramid: reading, collecting data, building models and turning them into slides.

The pyramid exists for a good reason. The work at the base is large, and someone has to do it. But every layer between the client's question and the person answering it adds a hand-off. The analyst summarises for the manager, the manager summarises for the partner, and the partner summarises for the board. Each summary loses a little of the evidence and adds a little distance.

By the end, the person who signs the recommendation can be the person furthest from the work behind it.

AI changes the base of the pyramid. Reading documents and public sources, tagging each claim with its source, drafting options and arguing against them are now work that AI agents do well and quickly. That is most of what the layers were for.

It does not change the top. AI does not remove the strategist. Someone still has to frame the question, decide what evidence counts, judge whether an option makes sense for this company, and put their name to the answer. Nor does the model do the arithmetic. The LLM explains; it does not calculate.2Show source for note 2 Every figure comes from our own deterministic code, so it can be tested and re-run.

What AI removes is the layers between the question and the person who answers it. That is a different claim from saying AI replaces strategists, and we think it is the more accurate one.

So we organise around pods, not pyramids. A strategy1 pod is 1 Strategy Partner and 1–2 Forward Deployed Strategists, supported by shared ontology and AI engineers.3Show source for note 3

• Strategy Partner. Owns the client's trust and signs the decision memo.
• Forward Deployed Strategists. Frame the problem, run the engines and work on-site with the client.
• Ontology engineers. Design and maintain the Strategy Twin, the living model every engine works from.
• AI engineers. Build the engines, the functions and the connectors.

Domain Fellows, who are chartered engineers, CAs, lawyers and sector veterans, sign what carries professional liability.4Show source for note 4 They are paid per review. A pod does not pretend to hold every licence. strategy1 provides decision support and blueprints; licensed professionals sign technical, financial and legal conclusions.

The engineers are shared across pods. The strategists are not. That is deliberate: the people who build the machinery serve many engagements, and the people who answer for a client's decision stay with that client.

A pod can carry work that once needed a pyramid because it does not carry it alone. Each engine combines a strategist pod, a team of AI agents, calculation code we test ourselves and a red team, all working from the client's Strategy Twin.5Show source for note 5

The agents do the reading and the drafting. The code does the arithmetic. The red team argues against every option before it reaches the Council, and writes down the conditions under which it fails. The Twin holds what has been learned, so the next engagement starts where the last one stopped instead of with a fresh round of research.

That changes the time the work takes. A pod engagement runs from 72 hours to 3 weeks.6Show source for note 6 It also changes how the work is priced. When the effort is not a pyramid of hours, there is no reason to sell it by the hour. We work on a fixed scope and a fixed fee, not hours.

The point of a pod is not that it is small. It is that the senior people are close to the work. The Forward Deployed Strategist who framed the question is in the room when the options are argued. The Strategy Partner who signs the memo has seen the evidence ledger, not a summary of a summary.

That closeness is what makes the gates work. In the S1 Method a named person signs at three gates: Model, Decide and Deploy.7Show source for note 7 An ontology engineer confirms the structure of the Twin. The Strategy Council and, where liability exists, a Domain Fellow sign the decision, and the client chooses. The client's named owner approves a dry-run before anything is written to their systems. A signature means more when the person signing knows what sits underneath it.

A pod asks something different of the people in it. There is no layer to hand work down to and no layer to hide behind. A Forward Deployed Strategist works with AI agents every day, for research, modelling and red-teaming, and decides whether what they produce is good enough. Every number they publish has a source and a grade. Every plan they ship says how it fails.

It also needs roles the pyramid never had. Eval engineers test that our agents are right, and measure how often. Function and quant engineers build the models behind the numbers. The pyramid did that work by hand, one spreadsheet at a time.

We are building this from India, for Indian companies first, then the GCC. The pod is how we intend to grow: not by adding layers, but by adding pods, each with a senior person whose name is on the decision. AI does the heavy lifting between the gates. People hold the gates. The pyramid was a way to organise the heavy lifting. When the lifting changes, the shape should change too.