A practical system for students and young architects who want clarity, not chaos.
Most architecture students don’t struggle with software. They struggle with decision-making. They can model, render, and produce diagrams, but when asked why a project works, the explanation becomes vague, emotional, or improvised. The real problem is rarely creativity — it’s the absence of structure.
This article introduces a simple but powerful system that helps you:
-
-
Define a clear design thesis.
-
Track and justify decisions.
-
Avoid emotional redesign cycles.
-
Detect logical inconsistencies early.
-
Use AI as a design reviewer rather than a design generator.
-
This is not about automating architecture. It is about strengthening your thinking.
The Hidden Problem in Most Student Projects
There is a pattern that repeats every semester: a “strong image” appears, the form evolves quickly, but the narrative is reconstructed only after the design is finished. The result often looks impressive, but it lacks coherence.
Common symptoms include:
-
-
A thesis that sounds abstract but doesn’t affect spatial decisions.
-
Diagrams that don’t match the built form.
-
A project that answers aesthetic questions but ignores brief conflicts.
-
Major changes triggered by a single critique comment.
-
Rendering starting before the spatial logic is resolved.
-
The Architectural Knowledge Base System
Create a folder dedicated to your current project. Inside it, create six simple text files. This is about externalizing your thinking so it can be tested.
| File Name | Purpose & Full Content Details |
| 1. project.md |
Your Holy Grail. Answers: Why does this project exist in this form? Write: The Problem, The Conflict, The Core Thesis, The Spatial Strategy, and The Main Risk. Example: Public Library – Berlin. Problem: Noise; Conflict: Openness vs acoustic control; Thesis: Acoustic gradient; Strategy: Public buffer; Risk: Reduced transparency. |
| 2. criteria.md |
Your Safety Algorithm. Your non-negotiable rules.
Example Rules: Form is never the first decision; every move must respond to context; no rendering before circulation is diagrammed; if I can’t explain it verbally, it doesn’t belong. |
| 3. watchlist.md |
Your Idea Radar. Not a mood board, but a list of ideas to test — but only under conditions.
Logic: |
| 4. decision_log.md |
Logs of Your Brain. Record major changes like a server log.
Example: |
| 5. lessons.md |
Post-Mortem. Every failed presentation is paid education. Write: What worked? What failed? What criticism repeated?.
Example: “Mistake: Developed façade before movement strategy. New rule: Diagram circulation before massing.”. |
| 6. context.md | The Real World. Notes on: Climate strategies, Regulatory constraints, Local precedents, Competition trends, and Social behavior patterns. This file explains “why” when a critique gets tough. |
Now We Connect AI
Instead of asking AI “What kind of concept should I use?”, you feed it your own thinking. Drag your 6 files into Claude, NotebookLM, or ChatGPT to begin a “code review” of your architecture.
5 Prompts You Can Use Right Now
| Goal | Prompt Content |
| 1. Coherence Audit | “Based on my project.md and criteria.md, evaluate whether my design decisions are logically consistent. Identify contradictions and weak assumptions. Be strict.” |
| 2. Red Teaming | “Read my thesis in project.md. Act as a strict competition juror. List the 5 strongest criticisms that could cause this project to fail.” |
| 3. Design Psychology | “Analyze my decision_log.md. Identify patterns in my design behavior. Do I change direction under stress? Do I overreact to feedback? What recurring cognitive biases do you see?” |
| 4. Brief Alignment | “Compare project.md with brief_breakdown.md. Am I actually solving the core conflict of the brief, or am I focusing on aesthetics?” |
| 5. Retrospective Loop | “Read lessons.md. Summarize the 3 recurring mistakes I keep repeating. Propose one strict rule I should add to criteria.md to prevent them.” |
Important: AI Is Not a Magic Wand
AI will not design for you, decide where your entrance goes, or tell you if you’ll win. It acts as a mirror. When you write your thesis and AI exposes its logical holes, you stop designing on vibes and start designing consciously.
Your 30-Minute MVP:
-
-
Create a folder:
MyProjectSystem. -
Create 3 files:
project.md,criteria.md,decision_log.md. -
Fill them roughly: 5 sentences for thesis, 3 rules for criteria, last 3 changes for log.
-
Upload to AI and ask: “Is my project coherent or am I improvising?”.
-
Homework (Do This Today):
Open project.md and write ONE sentence: Why did I design it this way?. If you get stuck, it means you’re designing on instinct, not structure. That is your point zero.
The post How to Use AI to Strengthen Your Architectural Design Process appeared first on Competitions.archi.