Foresight · The Agent Economy
Autonomous AI agents are about to do the work — buying, negotiating, researching, governing, deciding. That doesn't just add a tool; it rewrites the economy, the institutions and the balance of power. We do strategic foresight for that world: mapping how the agent economy reshapes markets, states and society — and what to do before it arrives, not after.
Reports · download
Our reports take one hard question about the agent economy and answer it in depth — what changes, who wins, what a government or institution should do now. Read it free online, or download the full report.
Four divergent futures for government and society once agents do much of the everyday work.
What we do
Most "AI foresight" is about using agents to do the analysis faster. That is the easy part, and there are tools for it. The hard part — the part that matters — is seeing what the agents do to the world.
An economy where autonomous agents transact, hire, research and decide is a different economy. It moves faster, concentrates differently, breaks in new places. We study that shift the way the world's best foresight shops study any deep transition: rigorously, from many angles, and far enough ahead that leaders can still act on it.
How we work
Foresight isn't guessing. It's a discipline. These are the six moves we run on every question — from scanning the first faint signals to handing leaders a decision they can actually make.
01
Scan
We read widely and constantly — research, patents, filings, markets, the fringes — to catch the first signs of how agents are changing work, money and power, while they are still weak.
02
Sense-make
We separate the noise from the forces that actually move the system — the structural drivers and assumptions beneath the headlines — so the analysis is about causes, not vibes.
03
Synthesise
We draw out several distinct, internally-consistent futures for the agent economy — not one forecast, but a span of how it could plausibly go, so no single bet goes unexamined.
04
Stress-test
We argue against every conclusion — inject shocks and wild cards, attack the load-bearing assumptions, and keep only what survives. If the thesis holds up, it is worth acting on.
05
Simulate
We model the second- and third-order effects — what a change sets off down the line — so the consequences are visible before a decision is taken, not discovered after.
06
Steer
We translate the futures into clear choices, early-warning indicators and concrete moves a leader can make now — foresight wired to action, not a report that sits on a shelf.
What we believe
01
Foresight becomes a continuous service, not a biennial workshop whose output ages on a shelf.
02
Agents map the is of what could happen; the ought stays with an accountable person.
03
Spin up disagreeing agents on purpose. A future with one author is propaganda, not foresight.
04
Every weak signal, scenario and claim is sourced, logged and auditable — back to the document it came from.
05
Tune agents to the periphery. By the time a trend is loud, it is no longer foresight.
06
Reward plausible surprise, not consensus comfort. The useful scenario is the one you didn't want.
07
Wire foresight into the choices it should change. Measure it by decisions altered, not pages produced.
08
A public capability the institution runs — not futures rented per-seat from a consultancy.
The canon, rebuilt
We didn't throw out the discipline's best tools — we gave each one a tireless workforce. The method on the left has half a century of practice behind it; the agentic form on the right is what it becomes when intelligence is abundant.
Horizon scanning
Hundreds of agents watching different corners of the world, surfacing anomalies a small team would never reach.
/method/horizon-scanning →
Scenario planning
Shell-grade scenarios produced and refreshed continuously, branching as new signals arrive.
/method/scenario-planning →
Delphi method
Diverse persona-agents debate to convergence in minutes — then real experts adjudicate the result.
/method/delphi-method →
Causal Layered Analysis
Machines drill from litany to system to worldview to myth, so strategy hits the deepest layer that moves.
/method/causal-layered-analysis →
Futures wheel
Agents fan out first-, second- and third-order effects into a navigable causal map, not a whiteboard sketch.
/method/futures-wheel →
Backcasting
Fix the preferred future, then let agents reverse-engineer the milestones, triggers and dependencies to reach it.
/method/backcasting →
Wild cards & weak signals
Outlier-seeking agents prize the low-probability, high-impact event the trend-followers systematically miss.
/method/wild-cards-weak-signals →
Cross-impact & morphology
Agents enumerate and prune the combinatorial space of drivers no human team has the hours to cover.
/method/cross-impact-morphological →
Where the method comes from
Foresight isn't something we invented. A handful of government units, think tanks and corporate teams have spent decades getting it right — running national horizon scans, building the scenarios that warned of oil shocks and pandemics, advising prime ministers. We study how each one works, take the methods that have proven out, and point them at one question: the agent economy.
See it coming
We help governments, institutions and companies understand what the agent economy does to them — and decide what to do about it while there is still time to choose.
