How do you turn years of transport research, modelling and real-world pilot experience into a tool that another city can actually use? That was the challenge behind the development of the OPTI-UP Toolkit.
The answer was not to simply place project reports online or compress hundreds of pages of research into a digital questionnaire. The aim was to translate the knowledge generated throughout OPTI-UP into a structured decision-support process that could help other small and medium-sized cities identify promising directions for improving their public transport. That process eventually became the OPTI-UP Toolkit.
The result was built progressively: from research and modelling, through real-world pilot experience and expert knowledge, to technical development and repeated testing.
Step 1: Start with real public transport challenges
OPTI-UP worked across six project areas in Central Europe: Osijek, Modena, Paks, Pécs, Český Krumlov and Grosuplje. The cities differ in size, transport networks, demand patterns and planning capacity. That diversity was important.
During the first part of the project, partners collected and analysed data on existing public transport networks and planning practices. Transport models were then developed or updated to better understand current conditions and test possible future scenarios.
The modelling work itself demonstrated why a simpler decision-support tool could be useful. Modelling capabilities varied considerably between project areas, while data availability and local technical expertise were recurring challenges.
Step 2: Test ideas in the real world
Analysis and modelling were not the end point. The project used this work to develop local plans and implement pilot actions covering three main areas: demand-responsive transport, public transport network optimisation and the use of alternative-fuel vehicles.
That meant the project could compare theoretical planning with implementation: what worked as expected, what changed once an intervention reached the real world and what practical challenges emerged.
The project’s comprehensive strategy was itself developed iteratively with partner cities, with methodological components reviewed and refined using stakeholder feedback and real-world application. These experiences became an important foundation for the Toolkit.
Step 3: Translate project knowledge into decision logic
The next challenge was considerably harder: turning all that knowledge into something a computer could use.
The development team first identified the city characteristics that matter when considering different public transport strategies. An initially broader set was gradually refined into 28 variables covering socio-demographics, public transport supply and demand, fleet characteristics and policy priorities.
At the same time, the project developed a bank of expert assessments representing different hypothetical city profiles and the strategies transport experts considered appropriate for them.
Step 4: Build logic that can work with imperfect information
A conventional rule-based system would require cities to fit predefined categories exactly. A machine-learning model, meanwhile, would require substantially more training data than the project had available. The team therefore adopted a fuzzy rule-interpolation approach.
In practical terms, this allows the Toolkit to work with approximate information and intermediate situations. A user can provide exact numbers where they are available but can also describe certain conditions qualitatively. The system then assesses similarities between the user’s city and the expert cases instead of looking only for an exact match. That logic forms the analytical core behind the recommendations users see today.
Step 5: Turn the logic into a usable digital product
Once the underlying methodology had been established, technical development translated it into a web application.
The architecture was organised around four stages:
- Collect city parameters.
- Rank potentially suitable strategies.
- Match the selected strategy with relevant OPTI-UP reference solutions.
- Help users interpret and explore the result through an AI-supported interface.
A Retrieval-Augmented Generation (RAG) pipeline was also introduced for the AI phase. Rather than relying only on the language model’s general knowledge, the system retrieves information from OPTI-UP documentation to support its answers.
Step 6: Test. Collect feedback. Improve. Repeat.
Testing was not left until the end. The Toolkit was evaluated iteratively throughout development. Three perspectives were deliberately included: transport experts and urban planners, IT specialists, and potential end users.
Transport experts examined the recommendation logic, terminology and parameter definitions. IT specialists focused on performance, system integration and technical stability. Potential users concentrated on something equally important: whether the Toolkit was actually understandable and easy to use.
The feedback was concrete. Questions needed clearer wording. Some parameters needed better explanations. The interface contained too much information in places. Navigation needed simplifying. The chatbot’s first response took too long. Introductory AI messages were too long. Users wanted clearer instructions and easier access to reference documents.
Those observations led directly to changes. Screen layouts were redesigned. Interface elements were removed or reorganised. Explanatory text was added. Question wording was revised. Expert parameters and rules were refined. AI prompts and the retrieval process were adjusted to produce clearer and more relevant responses.
From project result to practical tool
The Toolkit that users see today is therefore not simply the output of software development. It is the result of a longer chain: research → data → modelling → local plans → pilot experience → expert knowledge → decision logic → development → testing → refinement.
That process also explains the role the Toolkit is intended to play. It does not attempt to reproduce the complexity of a complete transport study inside a browser. Instead, it turns the accumulated experience of OPTI-UP into an accessible first step for cities that want to understand which public transport improvement pathways may deserve closer investigation.
The Toolkit may be digital, but the knowledge behind it came from real cities, real planning challenges and repeated testing of how public transport decisions work in practice.
See the result for yourself. The OPTI-UP Toolkit brings together the research, modelling, pilot experience and expert knowledge developed throughout the project in one interactive tool.
Try the OPTI-UP Toolkit