Secrets of Winning Architectural Proposals with AI

This was the exact opening line of my talk on stage at Al Hussein Cultural Center during the Young Architects Forum — with over 1,000 people in the audience.

I started by talking about the very first time humans imagined automation and the idea of artificial intelligence. That moment when we began to wonder: What if work could think for itself?

Then I shared 4 secrets that helped me turn AI into a serious tool for architectural proposals — on the same level as traditional modeling and rendering methods.


1. Scene Prompting, Not Project Prompting

The first mistake I see people make is jumping straight into writing prompts about the project as a whole.

Instead, write a scene prompt. Visualize that you’re holding a camera, capturing your project from the outside or inside. Is it an aerial shot of the site? A cozy corner of the living room? That’s your scene.

From there, describe what you see in text. This is where your role as an architect kicks in — turning vision into narrative.


2. Prompt Structure is a Skill You Train

In the beginning, your focus should be on how to sequence elements of the scene logically.

This habit helps you learn how to break down visuals into text, remember architectural details, and stay aligned with either the client’s brief or your own design logic.

In the lecture, I broke down the exact steps to follow. It’s a skill, and with practice, it becomes second nature.


3. Timing is Everything

This one is both creative and managerial.

The key is knowing when to use AI and why. If you jump too early or too late into AI tools, you might lose the proposal’s clarity or creative edge.

I spoke about design phases — and how understanding the goal of each phase helps you decide if AI should assist with massing, mood, structure, or rendering.

It’s not about throwing prompts and hoping for the best. It’s about timing AI like a professional tool.


4. What’s Next After Prompting?

Always ask yourself: What comes after mastering prompts?

Here’s where I introduced LoRA — Low Rank Adaptation.

I explained how I trained the AI model to understand our firm’s architectural language. Not just general architecture, but the way we think, design, and present.

This made our process way smoother when building proposals using AI. It felt like the system finally “got” our design logic.


By the end of the talk, I saw over 100 hands raised, waiting to ask questions. That was the best feeling. I knew the message had landed, and the energy in the room was something I’ll never forget.

This entire segment took just 25 minutes — but behind it was a full week of preparation.

Thanks to everyone who attended or tried to attend.

Instead, write a scene prompt. Visualize that you’re holding a camera, capturing your project from the outside or inside. Is it an aerial shot of the site? A cozy corner of the living room? That’s your prompt.

Saif Smeirat

Facebook Video for Reference: Facebook

Previous Article

Podcast: AI, Architecture & My Journey – with Saif Smeirat

Next Article

How I Designed My Own MBA: A Self-Study Blueprint from an Architect in Business

Write a Comment

Leave a Comment

Your email address will not be published. Required fields are marked *

Subscribe to my Newsletter

Subscribe to my email newsletter to get the latest Architecture, AI, and leadership insights posts delivered right to your email.
Pure inspiration, zero spam ✨