05 Your structural map
Reshuffle Index - Structural Foresight
Where are value, capability, and work moving in your domain - and which futures should you be planning against, with the evidence to defend the call?
Most work on the future of an industry delivers a study: a few themes, some archetypes, three scenarios, a recommendation. It is a snapshot, it ages the day it ships, and its claims cannot be tested. When the question is a national sector to steer, a multi-year capital allocation, or a workforce pipeline to build, a snapshot is not enough. You need a model you can interrogate, that names its own assumptions, that updates as the world moves, and whose every claim can be checked.
Don't buy a study. Stand up a living model - and run your decisions through it.
▍ A quick check
Early signs.
- You're being asked to take a defensible position on where a sector, market, or workforce is going - and “here's our point of view” won't survive scrutiny.
- The thinking in your domain feels superficial, and you can't tell which claims are structurally sound and which are extrapolation.
- You need to plan against more than one future, with the conditions that distinguish them named explicitly.
- The decision is multi-year, and a study that's stale on arrival is not good enough.
If two or more land, this is likely your question.
▍ The argument underneath
Why this matters.
If you read one section, read this. The argument the engagement is built on, from the ground up.
- 01
A domain has a structural logic that can be modelled
Every sector runs on an implicit logic: how its value chain is composed, where rent accrues, which positions are defensible. That logic exists independently of any one firm's strategy. Reconstructed from first principles as a typed model - stages, firm archetypes, roles, and the causal relationships between them - it becomes something you can reason over and stress-test, not just narrate.
- 02
Forecasts are only credible when they are falsifiable
A claim that cannot be wrong explains nothing. In this model every causal relationship names the observation that would refute it, and every forward claim traces a chain from cause to predicted outcome. That is what separates a defensible projection from a confident guess - and it is what lets a stakeholder interrogate the model rather than take it on faith.
- 03
Theory earns its keep by being checked against the record
Each relationship in the model is a hypothesis about how the domain works. A continuously maintained corpus - capability releases, funding, deals, regulation, real deployments - is read to see which hypotheses are firing and which are breaking. The theory proposes the structure; the unfolding record decides which parts are real. That is why the model stays current rather than frozen, and why its predictions are accountable to evidence rather than conviction.
▍ The work itself
Our work together.
5 phases. Each builds on the last - from analysis to a blueprint you can act on.
-
Scope the question
Fix the domain, the geography, and the decision the model has to serve. Lock the structural anchors - which value-chain stages, firm archetypes, and roles are in scope.
Deliverable A sealed scope: domain x geography x decision, with the structural anchors agreed.
-
Calibrate the baseline
Instantiate the structural model for the domain and geography against the corpus and the causal grammar. Establish today's baseline - how the structure is composed right now, every element traced to evidence.
Deliverable A calibrated baseline model of the domain's current structure.
-
Model the scenarios
Configure a small set of named futures - typically three constructive and one credible failure mode - each a distinct configuration of the forces in play. Project each forward through the model.
Deliverable A scenario set: for each future, which stages, archetypes, and roles strengthen, compress, collapse, or emerge, on what timeline, under what conditions.
-
Deliver and audit
Produce the calibrated report and, if commissioned, the live model the client runs scenarios through. Subject every claim to an independent cold audit before release.
Deliverable The report; optionally gated live access to the model - audited, traceable, defensible.
-
Maintain the model
The model keeps ingesting the live record. The client re-snapshots on a cadence; the picture sharpens as evidence accumulates.
Deliverable A model that stays current, plus a standing view of which structural bets the record is confirming or breaking.
▍ A recent engagement
Agentic competition across the customer journey
Agentic-behaviour modelling for competitive foresight
A consumer-facing business watched AI agents begin to intermediate its customer journey - starting at the top of the funnel, where buyers now open the journey inside a general assistant rather than on the company's own surface. Rather than a one-off competitive study, the engagement stood up a model of agentic behaviour across every stage of the journey: where agents already intervene, where they will as capability deepens, and how the intensity of competition shifts point by point. The model mapped new vectors of attack - the stages an entrant or an agent platform can insert itself and capture the relationship - against the stages that stay defensible because they bind a constraint agents do not dissolve. It separated the front of the journey, where agentic adoption is already deep, from the decisions closer to purchase, where it is not yet - and traced each claim to the adoption record rather than to assumption. From that, four competitive scenarios: assistant-led disintermediation, an agent-broker layer forming above the category, a defended direct relationship, and a price-comparison endgame that erodes margin.
▸ The shift
Outcome. A point-by-point read of which stages of the journey are contestable, which are defensible, and the conditions that decide each - plus a scenario set the client re-runs as agentic adoption moves down the funnel. The model surfaced two vectors of attack the client was not defending, and stays current as the competitive picture shifts rather than dating from the day it ships.
Let's discuss further.
If this question is in front of your team, the next step is a short call to scope it. Tell us what you're working through and we'll figure out together whether this is the right place to start.
▍ Or read first
Further reading.
Short essays from the Reshuffle series that build the argument behind this work.