THIRD ORBIT

Climate Adaptation AI

Climate adaptation failures are rarely a data problem. They are a coordination problem - between disciplines, between policy and practice, between the pace of scientific update and the cadence of institutional decision-making.

This project develops multi-agent AI systems designed to replicate that coordination at scale. Individual agents operate as domain specialists - each drawing on live APIs, continuously updated regulatory context, and authoritative scientific feeds. Orchestrated together, they function less like a tool and more like a standing technical advisory capability: available on demand, current by design, and able to synthesise across domains that rarely speak to one another in practice at this pace.

The work moves beyond static analysis toward systems that actively track shifting policy environments, flag emerging exposure, and support decision-makers with evidence that is both timely and tractable.

Active development. Selective collaborations considered.

Built Environment - AI Strategy

The built environment sector - architecture, urban planning, infrastructure consulting, construction - is structurally late to AI adoption, yet uniquely exposed to its consequences. Firms in this space face a compressed window to determine where AI creates genuine leverage and where it introduces risk to practice, reputation, and margin.

This work examines how specialist firms can develop durable AI strategies: not vendor-led, not experimental for its own sake, but grounded in the operational realities of project-based professional services. The question is not whether to adopt AI, but how to compete and build long-term advantage without losing what makes these firms irreplaceable.

Selective engagements. Enquiries by introduction.

Urban Co-Pilot - UNSW, Archistar, Pexa

In collaboration with researchers at the University of New South Wales, this project explores the design and evaluation of AI-assisted decision-support tools for urban planning and property practitioners. The work asks a precise question: when a planner or designer is mid-process - navigating trade-offs across community need, policy constraint, and spatial evidence - what does a genuinely useful AI co-pilot actually do, how does it work, and what barriers face adoption?

The focus is on augmentation, not automation. The research draws on human-computer interaction, planning theory, and emerging agentic AI capability to prototype tools that extend practitioner expertise rather than bypass it.

Active research collaboration. Details available on request.

Public Participation in the Age of AI

Democratic legitimacy in planning and infrastructure has always depended on meaningful public participation. AI changes the conditions for that participation - in how evidence is produced, how options are communicated, and increasingly, how community voices are synthesised and represented in decision-making.

This project builds on doctoral research in participatory analytics at UNSW and UCL to ask what participation should look like when AI is a substantive actor in the planning process. The work addresses both the methodological challenge - designing processes that remain genuinely inclusive - and the governance question of accountability when analytical outputs shape public outcomes.

Active development. Selective collaborations considered.