01
AI & intelligent systems
The strategic question is not whether AI matters. It is where performance and economics are sufficient to redesign the workflow - and where human judgement must remain accountable.
Patterns we recognise
→︎Demonstrations that do not survive deployment
→︎Productivity claims omitting integration and supervision
→︎Agents negotiating in your name, paid by the other side
→︎Governance designed apart from actual system behaviour
→︎Proprietary data exposed to models the business does not control
Decisions we resolve
→︎Which workflows justify investment
→︎Where margin lives once intelligence is commoditised
→︎Whether the harness - memory, context, proprietary process - stays yours
→︎Frontier API, open-weight or self-hosted - and where your data may flow
→︎Which judgements stay human - and at what cognitive debt
The record
Agents in production · c.250 senior hours saved in a quarter · apps, agent skills and public tools shipped end-to-end
02
Digital assets & trading
Digital assets, exchanges and payment rails share one question: which new infrastructure is genuinely faster or cheaper to run - and which revenue survives once incentives, liquidity and regulation settle. With the largest asset managers tokenising funds and US regulators admitting tokenised collateral, the question is no longer whether - it is where the economics are real, and who captures them.
Patterns we recognise
→︎Temporary market structure mistaken for durable revenue
→︎Market-maker liquidity or transaction volume mistaken for organic demand
→︎Tokenised assets carrying fewer rights than the assets they mirror
→︎Card or scheme growth extrapolated, not structurally sized
→︎Broker sentiment traded as if it were fundamentals
Decisions we resolve
→︎Where new payment or settlement rails are economically superior
→︎Whether tokenising your product wins distribution - or only re-plumbs cost
→︎Whether a marketplace has durable network effects
→︎How control and counterparty risk change the case
→︎Enter, invest, partner or wait
03
Consumer & retail
The data can be abundant while the demand logic is wrong. The answer often sits one segmentation below the category the industry uses - and one channel beyond the economics it reports.
Patterns we recognise
→︎Growth purchased after it stopped creating value
→︎External shocks left inside the run-rate
→︎Structural change extrapolated from historic behaviour
→︎Volume used as a proxy for customer value
Decisions we resolve
→︎Which customers and propositions deserve investment
→︎Where growth is genuinely incremental
→︎Which niches grow inside a flat category
→︎What scale and model the evidence supports
→︎Whether margins survive customers who shop by agent
04
Infrastructure & development
Large programmes and developments acquire momentum before demand, capacity and sequence have been independently established. The central risk is building the wrong asset at the wrong scale.
Patterns we recognise
→︎Demand inferred from the proposed capacity
→︎Prestige overriding commercial evidence
→︎Downside tested less hard than the central case
→︎Full scale presented as the only viable option
Decisions we resolve
→︎Demand independent of the proposal
→︎Capacity the market can support
→︎Components that create value
→︎Proceed, resize, resequence or stop