Gains:
- Ability to quickly analyze and summarize the structure of master zoning plan, implementation zoning plan and plan notes with the help of AI
- Ability to confirm each value from the officially approved plan while scanning the plan legend, function codes and construction conditions with AI
- Ability to use AI within its limits as a pre-review tool in plan modification, plan hierarchy and conflict analysis
The most intense reading task that lands on an urban planner's desk is deciphering development plans and their language. Planning in Türkiye progresses in a gradual hierarchy: at the top are regional and environmental plans, below are the master development plan (upper-scale plan that determines the general use decisions of the city, density and main transportation system at scales such as 1/5000), and at the bottom is the implementation development plan (lower-scale plan that determines the construction conditions on a parcel basis at a scale of 1/1000). These plans contain both drawing (legend, function colors, borders) and text (plan notes; written provisions describing how the plan will be implemented). AI is very helpful in quickly parsing, comparing and summarizing these long and technical texts. But every value he reads must be confirmed from the officially approved plan; because plan provisions are legally binding and misinterpretation jeopardizes the permit of a building or the law of a plan.
Plan hierarchy and why it is important
Plans are a chain: the lower-scale plan cannot violate the upper-scale plan. An implementation development plan cannot exceed the density and function given by the master plan. A planner's most common job is to check this harmony (or lack thereof). AI is very useful at juxtaposing two plan texts and flagging contradictions; but the planner makes the final decision with official documents.
Tip: When giving plan text to AI, be sure to specify what scale and date/approved plan it is. Context like “1/1000 implementation plan, approved 12.03.2023” frames the AI’s interpretation and makes it clear which document to look at when verifying.
Anatomy of plan notes
Plan notes generally include the following types of provisions: construction conditions (TAKS, KAKS/precedent, height), function definitions, drawing distances, parking and social equipment standards, special provisions (such as disaster, protection, coast) and general provisions. TAKS (Floor Area Coefficient) is the ratio of the area of the building sitting on the ground to the parcel; KAKS/precedent (Floor Area Coefficient) is the ratio of the total construction area to the parcel. These abbreviations are the core of the plan, and one misreading will ruin the project from the start.
AI can turn messy, unnumbered plan notes into a neat table, link referencing clauses, and produce a draft preliminary answer to the question "what provisions apply to this parcel?" However, AI sometimes hallucinates the item number, rate, or exception; so each line should be marked and confirmed from the original plan note.
Step by step: analyzing a plan with AI
- Identify the document. Note the scale, date of approval, plan type, and authority in charge.
- Anonymize text. Clear parcel/owner information if necessary.
- Remove the structure. Have AI tabulate the plan notes according to headings (structuring, function, equipment, special provisions).
- Mark the uncertainties. Request "CONFIRMATION REQUIRED" for each number and item.
- Hierarchy control. Have a conflict scan performed with the upper scale plan.
- Confirm. Verify each provision from the original approved plan; If in doubt, ask the competent administration.
- Leave your mark. Record which document confirms which provision.
Weak prompt / Strong prompt
Weak prompt: "What can I do with this parcel?"
AI produces a general estimate without the actual zoning status of the parcel, can make up rates and mislead you.
Powerful prompt: "Below is the text of the 1/1000 application development plan plan notes (approval date: [...]). Convert this text into a table: columns = Subject, Provision, Article no, Value. Put construction conditions, drawing distances, parking and special provisions on separate lines. Do not make up any value that is not clearly written in the text; write 'not found in the text' if it is unclear. Text: [...]"
The second prompt forces the AI to configure only what is in the text; It prevents fabrication and facilitates confirmation.
Four copyable templates
Task: Translate the following plan note text into a structured table.Columns: Subject | Judgment (summary) | Item/part number | Numeric value. Only use what is CLEARLY stated in the text; Mark what is not written as "not in the text". No proportions, distances or item numbers made up. Plan type/scale/approval date: [...]Text: [...]
Task: Compare the following two plan texts (upper scale and lower scale).List POSSIBLE contradictions in function, density and transportation decisions.For each contradiction: subject, upper plan provision, lower plan provision, "confirmation required".Decision making; just mark the highlights.Top plan: [...] | Subplot: [...]
Task: Explain all the abbreviations and technical terms (TAKS, KAKS/precedent, Yençok, towing distance, etc.) in the following plan notes in plain Turkish in a glossary table. Attributing value; just define the term.Text: [...]
Task: DRAFT an information memorandum to a citizen/stakeholder that explains the following plan provisions simply and without misleading. Give the technical values with the note "must be confirmed by the management". Do not claim legal certainty. Provisions: [...]
A table: types of plans and the role of AI
Plan/document
Content
Appropriate role of AI
Mandatory verification
Environmental layout / master plan
General use, density, main transportation
Text summary, comparison
Approved plan and legend
Implementation zoning plan
Construction on a parcel basis
Plan note tabulation
Approved 1/1000 plan
Plan notes
Written provisions
Configuration, conflict scanning
Original text, article by article
Zoning status certificate
Official information specific to the parcel
(Data entry/summary)
Authorized administration document
Plan amendment
Amendment proposal
Difference analysis draft
Suspension process, official decision
three mini cases
Case 1 — Confusion of precedent. A planner has the AI tabulate a jumbled plan memo. AI writes "precedent 1.60". The planner looks at the original text; the text reads "comparable 1.60, but 2.00 in commercial+residential mixed use". AI bypassed the exception. The planner fixes it. This shows that even the "only use what is in the text" rule is not enough in one pass, confirmation is a must.
Case 2 — Hierarchy paradox. A team compares an implementation plan and master plan to AI. AI marks when an area appears as "commercial" in the bottom plan and "non-residential urban work area" in the top plan. The planner takes this seriously, examines it with official documents, and finds a real incompatibility; He notices early that renovations are needed. The AI didn't make decisions, it just took notice; that's the value.
Case 3 — Citizen briefing. A municipality requires an information note that citizens can understand for a suspended plan. AI converts the technical plan notes into 1-page plain text and adds a "confirm from the administration" note for each numerical value. The planner reviews and publishes. A complex text becomes accessible in minutes rather than hours, without being misleading.
Common mistakes
- Missing exceptions. The words "however", "except", "special provisions" in the plan notes change the main rule; AI can bypass these.
- Not specifying scale and date. If you don't tell which plan you're looking at, the AI will interpret it from the wrong perspective; You don't know what to confirm either.
- To produce parcel-specific decisions. It is a serious mistake to have AI say "what to do on this parcel" and replace it with the official zoning status document.
- Mixing abbreviations. Confusing TAKS with KAKS, or precedent with height provision, spoils the entire project; Always verify terms.
- Mistaking contradiction scanning for a decision. The contradiction that AI flags is a hypothesis; The legal outcome requires official review.
In summary
Zoning and master plans consist of both drawings and legally binding text; plan notes are the core of the plan. AI is a powerful aid in structuring these long texts, explaining terms, flagging conflicts between plans, and producing citizen briefings. But AI can miss exceptions and hallucinate ratio/item. Each provision should be framed with scale and date of approval, confirmed item by item from the original approved plan; Parcel-specific decisions should be left to the official zoning status document. AI speeds up your reading of the plan, it does not approve the plan.
Application task
Take the text of a plan note in your hand (or a sample). (1) Have AI tabulated with the first template above. (2) Compare each row in the table to the original text and find at least one exception the AI missed or one detail it missed. (3) Create a glossary of terms with the third template and correct an incorrect/missing definition. (4) List which values you will confirm from which official document.
checklist
- [ ] I gave the plan type, scale, and approval date as context to the AI.
- [ ] I made the AI only use what's in the text; I prevented the fabrication.
- [ ] I also checked the exceptions and special provisions.
- [ ] I confirmed each numerical value and item number from the original approved plan.
- [ ] I left the parcel-specific decisions to the official zoning status document.
- [ ] I treated the conflict between plans scan as a hypothesis, not as a decision.
- [ ] I recorded which provision I confirmed from which document.