Our Research
To engineer autonomous enterprises, we connect three layers: how decisions are made, how agents coordinate, and how markets respond. Our work spans three scales, each informing the others.
-
Micro: How humans allocate attention and evaluate risk under uncertainty.
-
Meso: How teams of agents coordinate—negotiating resources, aligning goals, and executing workflows.
-
Macro: How strategies succeed or fail in simulated economies with competitors, consumers, and regulation.
AI Agents
The Meso Scale: Architecting the Distributed Company
We model an enterprise as a multi-agent system: specialised digital workers that coordinate like departments. Our research focuses on both agent design and the interaction layer, how agents communicate, negotiate tradeoffs, and align incentives. The goal is not “more automation,” but reliable coordination: systems that are auditable, robust to conflicts, and stable under changing conditions.
The Macro Scale:
Simulating Economic Survival
An autonomous enterprise should be tested before it’s trusted. We build agent-based simulations, synthetic economies populated by consumers, competitors, and regulators, to evaluate strategies across thousands of scenarios. These environments help us study emergent behaviour, market feedback loops, and rare high-impact shocks. We don’t aim to “predict” a single future; we evaluate many plausible futures to design systems that hold up.
Agent-Based Modelling
Cognitive & Decision Neuroscience
The Micro Scale:
Reverse-Engineering Decision-Making
Adaptation is the difference between a system that works in a demo and one that works in the wild. We study human decision-making under uncertainty and translate biological constraints into computational principles. Using methods such as high-density EEG, we quantify neural and behavioural signatures of attention, risk, and reward. These insights inform AI designs that are less brittle and better calibrated to real-world uncertainty.