Plan Guide

Municipality of Amsterdam

Unification and harmonisation of regulations for the physical environment to increase clarity and simplify implementation in the physical domain (VTH tasks and area development). In addition, it is a statutory task to complete the transition by 2032 and without deployment of the Plan Guide, this is not realistic.

Last change on 20th of April 2026, at 10:24 (CET) | Publication Standard 1.0

• Publication category: Other algorithms
• Impact assessment: Field not filled in.
• Status: In use

General information

Theme

Field not filled in.

Begin date

04-2026

End date

12-2031

Contact information

[e-mail]

Responsible use

Goal and impact

There is time until 31 December 2031 at the latest to convert more than 500 zoning plans into a single environment plan. To speed up the transition, with higher quality and with higher job satisfaction, an AI demand study was carried out. From this demand study, a number of scenarios were designed, including a total solution. This total solution addresses, among other things, the possibility of strengthening knowledge sharing and cooperation, enabling automation with regard to filling in conversion tables and ensuring consistency by reusing rules. A working prototype, called PlanGids, was realised and tested and demonstrated time savings, enthusiasm among colleagues to use the prototype and technical feasibility. A minimum viable product (MVP) was then developed that could be scaled up to all users after a pilot.

Considerations

Semantic search may inadvertently reproduce existing inequalities from old rules, and users may be inclined to adopt proposed rules more readily. In addition, differences in assessment between plan lawyers and lack of monitoring may lead to inconsistencies or undesirable patterns in outcomes.

Proportionality

The deployment of the algorithm offers clear advantages: it speeds up zoning plan conversion, promotes consistency by reusing rules and supports plan lawyers with targeted suggestions via semantic search. Disadvantages lie mainly in the risk of influencing human choices and unintentionally reproducing existing inequalities. As the system does not make autonomous decisions and the final substantive consideration always lies with the plan jurist (and subsequently the plan planner), the use of the algorithm is proportional and reasonably justified.

Measures

No decisions are made in the Plan Guide. Suggestions are still assessed outside the system by a plan judge. This is immediately a feedback moment when it can be checked whether the Plan Guide gives the right suggestions, or whether the Semantic Search needs to be adjusted.

Human intervention

Through the use of semantic search, suggestions are given to the plan jurist on how to translate zoning rules into environmental plan rules. The plan lawyer then decides whether one of the suggestions is appropriate for the existing rule. If this is not the case, then the plan lawyer can write a proposal for a new rule himself. This proposal is then reviewed by a plan judge, who then inserts the rule into another system. No decisions are made within PlanGuide.

PlanGids promotes equality by standardising rules, encouraging reuse and applying them consistently when converting to a single environment plan. Automation and cooperation reduce the influence of subjective interpretations and unconscious biases. In addition, transparency and traceability ensures that differences in treatment are understandable and correctable, thus countering unjustified discrimination.

Risk management

1. The PlanGuide uses only public data from zoning plans and the environment plan.
2. The Plan Guide is restricted by personal authorisation by role/function (only for people working on translations)
3. The PlanGuide only provides suggestions, no decisions are made.
4. A privacy quick scan has been carried out

Operations

Data

Field not filled in.

Technical design

The Plan Guide uses the public data of zoning plans. Only the zoning plan rules (ch 3) are loaded into a conversion table. In addition, the most recent version of the Zoning Plan is also loaded. Translated rules are stored in a Knowledge Base so that they can be used again.

Architecture of the model

1. Zoning plan rules (chapter 3 from zoning plan) are loaded in a conversion table.
2. The latest version of the Zoning Plan is loaded from public source.
3. Translated rules are stored in a Knowledge Base

Semantic search can be used to search per zoning plan rule in the Knowledge Base.

For each search result, the user sees to what extent it matches the search term (expressed as a percentage). Search results the finally chosen translation is logged and thus it is monitored whether adjustment of settings is necessary.

Link to code base

www.ruimtelijkeplannen.nl https://omgevingswet.overheid.nl/regels-op-de-kaart