AI+Design

Singapore World Expo Pavilion. 2036.

An AI-Integrated design journey exploring architecture, construction, business strategy and a pavilion designed for a second life.

The collection documents the complete design evolution through eight (8) assignments that was executed in 12 weeks, from initial research and conceptual development to architectural design, visualization, strategic thinking, and final presentation. Below is a brief summary of the succeeding projects, where I have labelled each step as a progression into the journey:

01 - Origin - Utilizing AI to Translating Sketch into a Story.
02 - Form - Using AI tools for the thought process and diagrams into the design.
03 - Logic - Utilizing agents to vibe code/ generate AI tools.
04 - Experience - Utilizing AI tools Adobe Firefly to enhance rendering.
05 - Documentation - Using AI plug-in for Revit to aid in drawing documentation.
06 - Story - Using AI tolls to generate a video to animate rendered images.
07 - Delivery - Utilizing AI tools to strategize a construction program.
08 - Position - Utilizing AI to formulate a business strategy / proposal for the Pavilion project.

The project began with a simple question: how can architecture demonstrate resilience in a permanently changing climate? Starting from hand sketches, the concept evolved around the anatomy of a mangrove/ tree-roots, trunk and canopy-as a metaphor for adaptation, growth and connection.

ChatGPT as the LLM and Run Diffusion Generative AI was used to help translate the initial idea into a broader architectural narrative and visual direction.

The project initialized in ChatGPT began with the question: What would be the most significant issue in the world in 2036? This conversation started several back-and forth ideation plus the sketch to come up with the baseline narrative for the entire project.

The concept was transformed into a repeatable architectural system. A 52' x 52' modular cell, flexible core, perimeter circulation, branching connections, environmental skin, and canopy established the rules that would govern how Rooted Resilience could grow horizontally and vertically. Rather than designing a fixed object, the pavilion became a system capable of adapting and expanding.

This chapter introduced two AI-driven strategies.
AI tool utilizing Anthropic Claude developed FITCHECK explored how AI could read an extensive fit-out criteria document and identify missing or potentially non-compliant requirements, while Growth Grammar translated the pavilion's design DNA into a parametric rule system that could generate and evaluate configurations against the project brief.

The focus shifted from AI as an image generator to AI as a design and coordination instrument.

A design does not exist in isolation—it is shaped by the people, environment, time, and activity around it. Using Adobe Photoshop and Firefly Generative Fill, the existing pavilion render was expanded into multiple contextual scenarios, exploring changes in people, planting, pavement, lighting, and time of day.

The exercise demonstrated how generative AI can extend and enrich an architectural visualization while preserving the original design intent.

The design moved from conceptual development into architectural documentation. Using Revit and SketchPro AI, the pavilion was translated into coordinated floor plans, elevations, schedules, room information, and presentation views.

This stage demonstrated how AI can support the transition from design intent to structured project information, while maintaining professional documentation standards.

The static pavilion was transformed into a cinematic experience. ChatGPT Image 3.0, Nano Banana 2, Kling 3.0 Pro and Adobe Express video editing were used to communicate the architecture through movement, light, sequence, and atmosphere—shifting the presentation from simply showing the building to revealing how it feels to move through it. The goal was to make the final animation feel curated as an architectural story rather than a software demonstration.

The project expanded beyond design into its complete physical lifecycle. The pavilion was developed as a temporary structure designed to be prefabricated, delivered, assembled, activated, disassembled, and reconfigured for a second life.
Anthropic Claude and Perplexity AI was utilized to analyze Construction strategy, mass timber, reversible connections, logistics, cost, risk, fire strategy, and reuse were incorporated into the design rather than treated as separate downstream problems.

The final stage translated the design into a client-facing business strategy.
The pavilion was positioned around five client priorities—visibility, certainty, cost control, sustainability, and legacy—and developed into a proposal centered on a pavilion designed for more than one life.
The project demonstrated how AI can support not only design and delivery, but also business development, positioning, and persuasive communication. Anthropic Claude was used to generate the proposal.