A city (or parts thereof) can be understood as a long⁃lived socio-technical system whose evolution is influenced by a lot of different interests and inter⁃dependencies:
（1）a spatial embedding of the plan object in question within a larger hierarchy (city, state, traffic system) with stronger specific relationships to neighboring regions, increasingly also in the third dimension;
（2）citizen or governance demands and resources from many different perspectives, e.g., mobility and logistics of goods/energy/water/data, industry and business development, sustainability, resilience, health, safety, and security;
（3）historical experiences and infrastructures, existing values created by earlier decisions or specific recent issues.
Therefore, a city can be also viewed as an entity of interaction within a complex network of goals and interdependencies. To disentangle this complexity, urban planning and management operate on at least two different time scales and across the three dimensions of structural, and dynamic properties of a city which are the fertile ground for implementing the citizens' requirements: 
（1）longer⁃term urban planning aiming from the viewpoint of a specific point in time for a defined future scope to create thew potential of a future infrastructure development with a defined time scope under goals including, e.g., attractiveness for citizens, business, tourism, but also sustainability including financial, ecological and capability aspects as well as adaptivity towards future ideas, innovations, and other opportunities, and resilience with respect to a wide range of relevant possible threats; 
（2）almost real⁃time operational planning, monitoring, and control of subsystems and their interaction with respect to important parameters based on IoT data management and related AI analytics.
Even if these two layers can to a large degree be studied independently from each other, there is also a need to study and manage their mutual interrelationships —— how statistics and issues observed at the operational level can be used, e.g., through operational data and the IoT to improve urban planning, or how knowledge captured at the urban planning level can be (directly but intelligently) reused for the operational tasks.
In order to capture all city planning aspects, there is the need to create models for the planning aspects at hand beyond maps and data warehouses. Digital Twins have already been successfully used in many individual urban planning tasks. AI⁃based digital twins and a related data exchange and simulation infrastructure facilitate modeling the real⁃world aspects and their interdependencies. 
The digital models can be implemented in the context of a broad framework of digital twin core conceptual models and services, which are defined by the international industry IoT consortium as:
"A digital twin is a virtual representation of real⁃world entities and processes, synchronized at a specified frequency and fidelity.
· Digital twin systems transform business by accelerating holistic understanding, optimal decision⁃making, and effective action.
· Digital twins use real⁃time and historical data to represent the past and present and simulate predicted futures.
· Digital twins are motivated by outcomes, tailored to use cases, powered by integration, built on data, guided by domain knowledge, and implemented in IT/OT systems."
The set of technological conceptual models called the digital twin core, focuses on interoperability aspects at the data and computational models, plus metadata, preprocessing, dataflow, and validation models. The report offers an impressive picture of the highly complex technological setting for Digital Twins from an object⁃oriented analysis and design perspective. 
This perspective has been augmented by the advent of large language models (LLMs) that already play a prominent role in urban planning. Recently, LLMs were generalized to Agentic AI systems which can play a powerful role in respect to multimodalities, knowledge acquisition, plan validation, simulation, and decision making. This new technology allows to create a framework for typical Urban Planning Agentic Social AI systems that integrates well⁃known Multi⁃Agent Systems technologies with Multimodal LLMs. This allows for a flexible, modular, and extendable approach in order to build a distributed cooperating and competing system of agents to capture the interdependencies of relevant urban aspects in a distributed Agentic Social AI System. 
For urban planning tasks, such an Agentic Social AI System is capable to model the relevant static urban structures, the necessary urban dynamics, and the rules, demands, and contributions of urban stakeholders. This also suggests an interactive workflow with feedback cycles including plan validations for the urban planning tasks at hand. This approach does not attempt to replace the human planner but instead provides intelligent support to the planners in a cooperative and interactive way based on multimodal communication, e.g., maps, images, reports, and voice. 
The goal of such a planning cycle is the development of an urban digital twin which provides the necessary static and dynamic plan details for a subsequent implementation and reflects a balance between the large set of requirements which must be satisfied because of existing urban planning knowledge, technological issues, administrative regulations, and the requests by virtual and later on, human urban stakeholders. This iterative planning process cycle entails, e.g., citizen requests, administrative regulations, financing restrictions, citizens' living and working conditions. Moreover, interdependencies between the many different urban planning aspects can be identified and consolidated, e.g., by balancing citizen demands with land use, roads, mobility, energy, water, parks, resilience properties, and population health aspects. To this end, the static structural urban plan elements should be completed first as a basis to subsequently incorporate the dynamic plan elements for process-oriented plan validation and simulation.
In a next step, this preliminary urban digital twin can be validated and improved by input from LLM⁃based stakeholder models implemented by argumentative digital twins based on specialized multimodal agentic AI models, that can bring specific, and even antagonistic aspects into the urban plan creation and negotiation using also human behavioral models. Therefore, expensive and time⁃consuming citizen planning contributions can be considered at later plan stages when "real humans" will get the opportunity for plan critique.
Moreover, these enabling (simulation-based) evaluation of the "operational feasibility" of urban plans by evaluating different IoT architectural options such as data distribution and governance, scalability and robustness/resilience/openness to future innovation analysis of an operational execution of the urban plan; addressing also semantic issues in enabling cooperation or competition of very large number of Digital Twins in an operational setting.
It is obvious, that the notion of urban digital Twins that are implemented by a society of multi⁃modal LLM⁃based agents ("Agentic Social AI System") extends the original definition of "digital twin" in respect of interoperability and complexity. However, given the complexity of cities and consequently, urban planning, the urban planning field will need to develop a framework along the lines outlined above in order to successfully cope with the planning complexity for livable cities. Moreover, such an urban digital twin can be used also as a framework for dealing with the operational requirements of real⁃time operational planning, monitoring, and control of urban subsystems and their interactions.