A personal reflection on why this company now exists, and where we are heading.
For more than twenty years, I have worked at the intersection of transport data, modelling and decision-making. My career began in 2003, straight out of university, building traditional four-stage models — tools that, whilst theoretically imperfect, remain remarkably effective at approximating real-world impacts.
One of the projects I am still proud of is the Region of Waterloo LRT work I contributed to in Canada. Years later, seeing the ION rapid transit thrive is oddly comforting: a reminder that even the humble four-stage trip-based model, handled with care (and under excellent guidance from mentors like Dave Crowley, Jim Chim and Karoly Krajczar), can stand the test of time.
Since 2012, most of my career has been in the UK — home to one of the most sophisticated appraisal frameworks in the world. I have been lucky enough to work on large national models, rail systems, public-transport modelling, bus analytics and more recently, intense investment-grade Traffic & Revenue (T&R) due-diligence work. That last piece was incredibly eye-opening: commercially sensitive, complex, high-stake, high-pressure and requiring absolute diligence. It also reminded me of an important truth — advanced modelling techniques are not always the best tools for managing risk.
This year, I finally had a chance to pause and think carefully about the future. And so, Amodellis Ltd. was born — not out of a dramatic reinvention, but out of a desire to step back, reassess where the field is heading, and contribute meaningfully to that direction.
We are well past the deep learning hype phase and now firmly in the era of agentic AI. Behavioural models remain essential, but it is naïve to think transport modelling can remain unchanged whilst every other analytical discipline transforms.
Recently, a thought-provoking post from Seb Krier caught my attention, highlighting the rise of multi-agent systems. The future will not be “one clever model for everyone”, but ecosystems of specialised agents modelling cooperation, competition, social dynamics, and spatial interactions.
This future feels distant — but the time to prepare is now.
Not too long ago, during a bid, I was asked to forecast demand for a TNC platform. We could estimate typical daily trips - but we also knew the real drivers were things our models barely touched:
Similarly, key future issues — such as EV charging constraints, autonomous vehicle fleet operations, shared mobility integration, and dynamic pricing mechanisms — sit entirely outside traditional TAG-compliant frameworks.
The world is changing faster than our models.
The DfT’s more proactive move toward an activity-based modelling (ABM) framework recently is significant and overdue.
As a behavioural modeller, I am comfortable with the core ingredients — choice models, segmentation, utility theory. But full ABM/agent-based systems require a larger ecosystem: simulation engines, state machines, temporal dynamics, feedback loops, and ongoing research culture that many consultancies struggle to sustain.
There is room for new players — especially those focused on behavioural realism, transparency and modern software practice.
Amodellis will be a transport modelling consultancy, yes — but one grounded in research, open tooling and modern workflows. Our vision is incremental and pragmatic:
This is why we have introduced a new service area — AI-Enabled Transport Research — as part of our long-term strategy. It sits alongside our core services in VDM, PT modelling, T&R forecasting, data analytics and choice modelling.
We want to contribute to the future whilst still supporting the present.
Founding this company is not about stepping away from modelling — it’s about stepping toward the version of modelling we want to spend the next decade working on.
One that is transparent, open, collaborative, and increasingly intelligent.
If you are an academic, a consultant, a public-sector modeller, a data scientist, or simply someone interested in improving how we understand and forecast human movement — we would love to connect. Let’s explore how we can build the next generation of modelling tools, together.
Amodellis begins small, but with purpose.
This is day one.
Here are some more articles you might like to read next: