Promte has presented a building permit app based on an analysis agent in connection with KL's large-scale AI project for building permit processing. The video above is the full KL webinar, where Promte presented alongside several other suppliers. The player starts at Promte's segment, where Victor S. Skytte, co-founder at Promte, walks through how the solution can help create a clearer overview of building permit cases, both before an application is submitted and when the case is later processed by the municipality.
KL describes the initiative as a project intended to strengthen municipal building permit processing through the use of artificial intelligence. The project is part of the agreement between the Danish government, KL and Danish Regions on shared large-scale AI projects.
The starting point is a challenge many municipalities know well: building permit processing often requires knowledge from many different sources. This can include application material, drawings, the building regulations, local plans, municipal planning frameworks, BBR information, internal workflows and previous practice.
When the relevant knowledge is spread across several places, asking a single question to a standard AI assistant is rarely enough.
Standard AI assistants work well when the user has a relatively precise question and the answer can be found in a clear source. That might be a question about a specific section of the building regulations.
Building permit cases are often more open-ended. The question is not always defined in advance, and the most important information may not be located in one specific place.
That is why Promte has worked with an analysis agent. It does not simply start by answering. It first tries to understand the context of the case: what type of case it is, which property it concerns, and which information is needed to provide a qualified overview.
The agent then defines a strategy for which sources should be examined. Along the way, the strategy can change if the analysis shows that additional knowledge is needed from other sources. Finally, findings, uncertainties and next steps are gathered in a summary.
This makes the solution better suited to tasks where the work is not just about finding one answer, but about creating an overview across complex material.
The presented solution has two entry points built on the same analysis agent.
The first is citizen-facing. Here, a citizen or company can upload material as a file or a folder with multiple files. Promte analyses the application and gives an overview of issues the user should be aware of before submitting the application.
The solution marks documents and checkpoints in red, yellow or green. Red means something is clearly missing. Yellow means the user should review something more closely. Green means that it appears to be fine.
The purpose is not to decide the case in advance. The purpose is to help applicants identify errors and missing material earlier, so the quality of the application improves before the municipality receives it.
The second entry point is aimed at building permit caseworkers. Here, the case can be added to Promte through integrations with relevant systems or by manually dragging documents into the app. The analysis agent reviews the material, searches across sources and highlights information that may be important for the further assessment.
This can include basic case information, address, local plan, municipal planning framework, zoning status and relevant parts of the building regulations.
An important point in the presentation was that the solution is not intended as fully automated building permit processing.
The caseworker still assesses the case, considers the professional issues and makes the decisions. The role of the analysis agent is to prepare the material, gather relevant information and make it easier to get started with the assessment.
That is also why the colour markings are used as checkpoints, not decisions. They help the caseworker quickly see where attention is needed, but they do not replace the professional assessment.
In the same way, the solution can prepare communication with the citizen. If the analysis shows that material is missing, the next step might be a draft letter requesting additional information. The draft must then be reviewed and adjusted by the caseworker before it is sent.
One advantage of building the solution on Promte is that it becomes part of a platform where much of the infrastructure is already in place.
Promte is used today by municipalities, agencies and private companies, and the platform has been developed in close collaboration with Danish municipalities. As a principle, new functionality is made available to all municipal customers at the same time.
The same applies to sharing. If a municipality develops a solution or workflow that works well, it can be shared through Promte's shared municipal library, if the municipality wants to share it. In that way, useful solutions can benefit more organisations without each municipality having to start from scratch.
For building permit processing, this means that the standard solution can be adapted to concrete municipal workflows. Promte contributes the platform, analysis agent and technical foundation, while municipalities contribute professional knowledge, local processes and experience from day-to-day case processing.
The building permit app shows how AI can become more practical when it is connected to a concrete workflow.
It is not only about chatting with a set of rules. It is about bringing documents, planning information, regulations and internal workflows into a process that both citizens and caseworkers can use to get a better overview.
When data ingestion, access, knowledge sources and operations already live in the Promte platform, the path from idea to concrete solution becomes shorter. It makes it possible to work more iteratively: start with a standard solution, test it in real workflows and adapt it together with the municipalities that will use it.
That is the type of AI use we believe most in for practical work. Not AI as a replacement for professional judgement, but as a tool that can prepare, structure and qualify the work, so employees can use their time better.
If you would like to hear more about how Promte can support building permit processing or other municipal workflows with AI, you are very welcome to contact Promte.
