How to Use AI for Architectural Programming

Let’s face it — with today’s complex building projects, architectural programming can be a real headache. When all said and done, the programming phase of architectural design can be very time-consuming — but with the help of AI, architects and designers can push designs to entirely new levels.

Not to be confused with “computer architecture” or “software programming”, architectural programming revolves around the architectural design process — the creation of physical spaces for people to inhabit. In fact, architectural programming the first part of the 6 phases of the design process for architecture.

Typically, architectural programming is thought to look like this:

However, there’s a lot more that goes into creating an architectural program than just creating bubble diagrams to show connections and space adjacencies.

Programming in architecture requires a lot of writing and number-crunching. This data is collected from previous architectural projects, technological research, client goals and expectations, and building codes among other things. Architectural programming is part of the pre-design phase of the design process, and acquiring accurate data is vital to a project’s success.

Once this data has been collected, architects can begin to distill the information into conceptual designs in the second phase of the design process, schematic design.

What I love most about architectural programming is the idea of “planting the seeds” for the design. Quite literally, an architectural program document becomes a “playbook” for the design. The architectural program will likely be a length document, regardless of the type of building being designed.

As you may expect, this “playbook” can take a lot of time and research to put together — and the way in which the programming data is collected is crucial for the project moving forward into later phases of the design process.

So, how can architects streamline the programming process? That’s where AI can help!

By using artificial intelligence, architects can streamline the programming process and gain deeper insights into their clients’ needs and preferences.

Using AI to save time in architectural programming can open up opportunities for focusing on other critical parts of the design process, and potentially create a better “playbook” for the design.

In this post, I’ll walk you through how you can use the latest AI software to help generate, write, conceptualize, and render content for an architectural program document.

As always, let’s start with the basics.

What is Architectural Programming?

Architectural Programming can be defined as the research and decision-making process that identifies the scope of work to be designed. It involves defining the requirements and objectives of a project, identifying the stakeholders, and creating a roadmap for the design process.

As the designer works closely with the client, important variables must be considered to ensure that the clients needs are met. Some of these variables may include, but are not limited:

• Who will the building ultimately serve?
• What program elements will the building contain?
• Where will these program elements be located?
• How will the building integrate within the site?
• What is the total size of the building?

Beyond the technical scope of programming, the designer must also consider experiential qualities, such as:

• What type of light will illuminate the space?
• How will the space deal with sound?
• What kind of materials will be used within the space?
• Which programmatic elements will be adjacent to one another?

If you want an example of an architectural program, take a look at this PDF:

Architectural Programming Example

Open PDF →

You’ll notice that the document is fairly lengthy at 118 pages. As I mentioned, it’s the “playbook” for the architectural design — and every design decision we make moving forward will be based on this document.

100+ pages may seem like a huge undertaking to write about a design that doesn’t actually exist yet — but I want to share some insights on how to use AI for architectural programming to help save time, acquire more accurate data, and ultimately create a better product.

How to Use AI for Architectural Programming

Here’s a “tool kit” of all of the AI programs for architectural programming that we’ll be looking at.

Architectural Programming AI Software

1. ChatGPT — Natural Language Processing (NLP) AI
2. Veras — Prompt-to-Image AI, Revit Plugin
3. Spacemaker AI — Building Site Programming AI
4. Aurivus — Real-world Site Scanning to 3D Digital Model Software

ChatGPT

Veras

Spacemaker

Aurivus

As for more “standard” design software for architectural programming, you’ll want to grab these:

1. Adobe InDesign — Document Creation
2. Adobe Illustrator — Vector Graphics Software
3. Adobe Photoshop — Raster Graphics Software
4. Revit — 3D Modeling Software
5. Sketchup — 3D Modeling Software

As for the 3D modeling software, it ultimately comes down to preference. However, at the time of writing this, the AI software that we’ll be using is tailored to use with Revit — so I strongly encourage you to use Revit.

Autodesk products are typically free for student use, otherwise you would need to purchase a license to use it. Note that Revit LT does not allow the use of plugins, so try to get a student version or licensed version of Revit if you can.

At its core, an architectural program will require architects and designers to generate a significant amount of content in the form of writing and graphics. Here are some pages from the sample architectural programming document above that we’ll be recreating in this walkthrough using AI:

Here’s a basic outline of an architectural program.

Basic Architectural Programming Outline

1. Project Description
2. Precedent Research / Gather Data
3. Site Analysis and Diagrams
4. Project Program Requirements
5. Program Analysis and Diagrams

This list could certainly expand depending on the type of project being designed, but this outline covers the basic information needed to create an architectural program.

Step 1: Create a Project Description

The first step in creating an architectural programming outline with AI is to define the objectives of the project. This includes identifying the goals of the stakeholders, the functional requirements of the building, and the budget constraints. This information can be gathered through surveys, interviews, and focus groups with the stakeholders.

Here’s how I would break down a project description for architectural programming:

I. Introduction

• Overview of the project and its purpose
• Explanation of the need for the project
• Brief description of the project site and context

II. Project Scope

• Description of the project’s goals and objectives
• Overview of the building program and functional requirements
• List of key stakeholders and their roles in the project
• Identification of any constraints or limitations

III. Project Details

• Description of the project site and its characteristics
• Analysis of the site’s opportunities and constraints
• Outline of the design approach and key design principles
• Description of the proposed building materials and systems

IV. Project Management

• Description of the project team and their roles and responsibilities
• Overview of the project timeline, milestones, and deadlines
• Budget and resource allocation
• Risk management plan and contingency measures

This outline can be tailored to fit the specific needs of each architectural project, but it provides a basic structure for organizing key information and communicating the project’s goals and requirements.

Using this outline as a basis, you can start to fill out each section manually based on the information that you have — which will take some time.

But to streamline that process, we can use the assistance of NLP (natural language processing) AI!

Use ChatGPT to Create an Architectural Project Description

First, open up ChatGPT.

Next, copy the following prompt and paste it into ChatGPT.

Write an overview of an architecture project and it’s purpose. Explain the need for the project and how it will serve the client and local community. Create a brief description of the project site and it’s context:

Following that description in the same text box, you’ll need to briefly describe the project with any data the you have. For example, here’s what I used as an input:

The project is a 300 key hotel in Texas City, Texas. It will be near the Gulf of Mexico, and the site is approximately 4 acres. The brand of the hotel cannot be disclosed, so we will refer to it as the “Texas City Resort” project.

Now, hit enter and let ChatGPT generate a description. Here’s the example result from ChatGPT:

Not too bad, right?

Obviously, we’ll need to fill out more specific details about the project. But as a starting point, we have some basic information that we can start to use to formulate the architectural program project description.

Thanks to the help of AI, it only took a matter of seconds to write 260 words! The project description can change as we work through the entire architectural programming process, but this is good enough to get started.

Next, we’ll need to integrate some architectural precedent research — which will actually inform more details to include in the project description.

Step 2: Precedent Research / Gathering Data

Precedent research in architecture refers to the study of existing buildings, structures, and designs to gather information, inspiration, and ideas that can inform the design process for a new project. This type of research involves examining the work of other architects, analyzing the design decisions that were made, and understanding the various factors that influenced those decisions.

The goal of precedent research is to build upon the knowledge and experience of others to create a new design that is informed by the best practices of the past and that addresses the specific needs and requirements of the project at hand. It’s an important tool for architects to use in order to develop a design that is grounded in the context of the surrounding environment, that responds to the needs of the user, and that is informed by the latest advancements in technology and materials.

Architectural Precedent Research and AI

How can architects use AI for precedent research?

One word. Pinterest.

Now, before you roll your eyes to the back of your head… did you know that Pinterest actually uses AI and machine learning for its image search algorithm?

That’s right, Pinterest has been using AI and machine learning in their visual search algorithm for over half a decade now. Here’s how:

“To properly label all photos uploaded, Pinterest deploys an AI algorithm that categorizes and sorts all photos. This makes it easier for images to be more efficiently ranked and recommended later in the process.”

Source: How does Pinterest use machine learning (ML)?

Pinterest employs representation learning, a type of deep learning algorithm that resembles neural networks, to categorize and label the massive number of photos uploaded to its platform. The efficacy of representation learning algorithms lies in their capability to extract highly specific details from images.

In fact, Pinterest started understanding the importance of AI back in 2015 and acquired a machine learning startup called Kosei that helps with content discovery and provides Pinterest users with highly personalized product recommendations.

Check out this article if you want to learn more about how Pinterest uses AI.

One of the more subtle features that Pinterest offers is the “Visual Search” tool — and if you don’t look closely enough, you might miss it.

In the bottom corner of each Pinterest image, a small magnifying glass allows users to select parts of an image to find other relevant and relatable images. This is how Pinterest implements AI.

Check it out, here:

Once you select the visual search button, you’ll be able to select specific parts of the image to search. In the results, you’ll be able to see other ideas that people have uploaded to Pinterest that have similar graphics.

For the use of architectural precedents, Pinterest AI helps streamline the process for collecting architectural precedents.

The only drawback with using Pinterest is that Pins can link to any website on the internet… so the source of the image may not be the original.

Just be sure to fact-check, and check all of the sources of the images to ensure that they’re coming from legitimate and useful websites.

Step 3: Site Analysis

When conducting a site analysis, it’s important to look at things like the terrain, climate, natural features, and nearby buildings. For example, if you’re building in an area with a lot of wind, you might want to design the building to be more aerodynamic. Or, if there’s a beautiful view nearby, you might want to make sure that the building has windows or other features that can take advantage of it.

Another important aspect of site analysis is studying the local culture and history of the area. This can help you design a building that fits in with the surroundings and feels like it belongs there. For example, if you’re building in a historic district, you might want to incorporate design elements that are similar to other buildings in the area.

Architects are getting pretty tech-savvy these days and are starting to use more advanced methods for studying potential building sites. Beyond the tradition uses of capturing photographs and sketching site conditions, the use of drones for site analysis is becoming ever more common.

Taking site photography to a whole new level, architects can use scanning software to capture highly-detailed information of a site that can be converted into 3D models to use for architectural programming. Site scanning helps to saved a immense amount of time, and can also help with design accuracy.

Here, check out one of the best AI site scanning software available for the AEC industry called Aurivus.

Aurivus AI for Site Analysis

Aurivus

Aurivus focuses on the “I” in BIM (Building Information Modeling), and “makes scan-to-BIM faster by assigning BIM attributes directly to the point cloud of the objects. Modelers can use it to reach their goal much faster.”

BIM doesn’t usually show up in the architectural design process during architectural programming, however tools like Aurivus AI are helping to streamline the integration process at the very beginning. During a site analysis, a user would scan an existing site and the information would be uploaded to the Aurivus AI point cloud.

That data could then be accessed on the design-side, and software like Revit could take the scan data and directly integrate it into a 3D model. Here’s an example of how this process works:

Pretty amazing, right?

Of course in order to accomplish this, you’ll need a LiDAR camera — and they don’t come cheap. The Leica BLK2GO shown in this example would set you back a cool $50,000+ (but that’s on the high-end of these devices).

However, you may be surprised to learn that you have a LiDAR capable device sitting in your pocket. Yes, your cellphone! In fact, Aurivus is making this technology more accessible by implementing its AI with Apple products, like iPhones and iPads.

Take a look at this video to see it in action.

Aurivus is unparalleled for capturing real-world site conditions using AI. But what if we don’t have access to this kind of technology? What if we can’t conduct a site visit for some reason?

Next, we’ll take a look at a program that’s still pretty fresh on the scene, but its future looks very promising.

Geopipe AI

To take site analysis into the digital world, software like Geopipe AI uses machine learning to create 3D models of real-life cities. They offer “free, instant real-world models” that can be used for gaming simulation or AEC (architecture, engineering and construction) purposes.

On the Geopipe website, they allow a user to “Construct a simulation or experience in a real city in an hour, not a year. Whether you’re an indie game developer, a AAA studio, an architect, or a VR artist, Geopipe gives you the tools you need to immerse your users in realistic environments.”

According to their website, it works like this:

“Geopipe’s AI starts with raw sensor data like photos, laser scans, and more. Our proprietary machine learning pipeline identifies every object in a city, then rapidly reconstructs rich digital twins at multiple levels of detail. We reconstruct regions ahead of time so you can download models instantly.”

As of writing this, you can head over to their “Downloads” page and grab a free model of Rockefeller Plaza and Columbus Circle in New York City.

After looking through the models that they’re offering, it’s pretty incredible that you can get this amount of detailed information free of charge.

Granted, the intent of these models is to serve as a base — and you’ll need to use other 3D software to populate the models with folliage, entourage, etc. But as a starting point, having these models could be quite useful for conducting a few quick studies for a site analysis.

Overall, site analysis is really important for the architectural programming process because it helps architects understand the context in which they are designing.

By integrating AI into the site analysis and diagramming process, architects can create more accurate and true-to-life representations of site conditions. AI is truly helping to bridge the gap between studying real-world and digital environments.

Step 4: Project Program Requirements

Project Program Requirements are basically a set of guidelines that architects use during the architectural programming process to make sure that the building design meets the needs of its users. These requirements are a vital part of the planning stage and are usually developed in collaboration with the building owner or client.

The requirements typically include a detailed list of functional spaces and areas that the building needs to have, as well as the dimensions and capacity for each space. For example, a hospital might require a certain number of operating rooms, patient rooms, and administrative offices, each with specific equipment and square footage requirements.

The requirements also consider the building’s environmental and technical needs, such as heating and cooling, ventilation, and lighting. They may even include guidelines for the building’s sustainability and energy efficiency goals.

Ideally, we’re trying to create something like this:

This is a theoretical program that I wrote about in this post, if you’re curious.

By defining the Project Program Requirements at the beginning of the design process, architects can ensure that the final building design meets the needs of its users and functions properly. It also helps to avoid any costly redesigns or construction delays later on in the process.

So, how can AI be used to create a document for Project Program Requirements?

ChatGPT for Architectural Programming

Here’s how I used ChatGPT to create a basic program outline for a 300 key hotel in Texas City, Texas.

This is the prompt that I used for ChatGPT:

“Create an architecture program with square footage calculations for a hotel with 300 keys on 2 acres of land by the Gulf of Mexico Coast. Include information on best materials for construction, estimated building occupancies, and applicable building codes for Texas City, Texas.”

Based on my experience in Hospitality and hotel design, the results are actually quite good. Take a look:

The “Estimated Building Occupancies” and “Building Codes” sections could certainly use some refinement, but this is a good basis to start with. In just a matter of a few seconds, ChatGPT has provided some basic information to start informing the architectural program.

Once we have this program information, the real fun can begin!

We’ll need to take the program data and create visual studies in the form of architectural design diagrams to understand how the architectural program functions.

Step 5: Architectural Program Diagrams

There are several types of architectural diagrams used for programming and pre-design. These diagrams are used to help architects and stakeholders understand the goals and requirements of a building project and to inform the schematic design of the space.

Here are some of the most common types:

Bubble Diagrams

Bubble diagrams are simple diagrams that use circles or bubbles to represent different spaces and functions. They are used to show the relationships between spaces and to identify adjacencies and groupings of spaces.

Block Diagrams

Block diagrams are more detailed than bubble diagrams and include information about the size, location, and relationship of different spaces. They are used to show the overall organization of the building and to identify any potential issues with space allocation.

Site Analysis Diagrams

Site analysis diagrams are used to analyze the site and to identify any constraints or opportunities that might affect the design of the building. These diagrams might include information about the topography, vegetation, views, sun angles, and wind patterns of the site.

Circulation Diagrams

Circulation diagrams show the movement of people and resources through the building. They are used to identify the most efficient and effective circulation patterns and to ensure that the building is accessible to all users.

Programmatic Flow Diagrams

Programmatic flow diagrams show the flow of activities and functions through the building. They are used to identify the most effective layout for the building and to ensure that all programmatic requirements are met.

Zoning Diagrams

Zoning diagrams show how different areas of the building will be used and how they will be separated from each other. They are used to ensure that the building meets zoning and code requirements.

All of these diagrams are used during the programming and pre-design phases of a building project to inform the design process and to ensure that the building meets the needs of its users.

How to Use AI to Create Architectural Diagrams

I created all of the diagrams in the example above using a mix of Revit, Illustrator, and Photoshop. These types of diagrams take quite a long time to make from scratch — but AI can certainly help expedite the process.

Using text-to-art AI generators, such as MidJourney or Night Cafe would create a lot of “random” diagrams that wouldn’t really prove to be useful for a specific design. At this point of the architectural programming exercise, we’re trying to take the data the we gathered in the beginning and turn it into easy-to-read graphics.

Since architectural programming diagrams are tied so closely to numeric data, we can’t introduce any bit of “randomness” into the process if we’re using these diagrams as tools to understand the framework of the design.

However, AI software that integrates with 3D software can help eliminate a lot of that “randomness” that is present in text-to-art generators. So, what software can do that?

Spacemaker AI.

Spacemaker AI for Architecture Test Fits

SpaceMaker

Typically, the next course of action in the architectural programming process would be to start generating test fits and diagrams of potential building layouts within a site. Spacemaker AI helps with that — and architects can produce design iterations in 3D faster than ever.

Acquired by Autodesk in 2020, Spacemaker is a generative design software that can be used for rapid digital prototyping in architectural programming. For site analysis specifically, Autodesk claims “Spacemaker’s analyses enable you to accurately test and understand your proposals during the early stages, planning in greater detail than ever before, especially when it comes to buildability, livability and compliance. They’re fast and easy to use, meaning you don’t need in-depth technical expertise to run an analysis.”

Source: Autodesk.com

The software allows architects and designers to conduct building analysis studies in 3D, while calculating relevant metrics related to the number of buildings on a site. For example, using Spacemaker for building analysis can provide information for the total number of buildings, average number of stories, total area of all facades, total volume, number of exterior corners, average building height, and maximum building height.

Here’s a screenshot from Autodesk’s website:

Personally, I find generative design especially helpful for architectural programming. By inputting different parameters for things like room sizes, functionality, and energy use, I can use generative design to come up with a bunch of different design options that meet all the requirements. It’s a huge time-saver, and it allows us to focus on the details and fine-tuning the best design option for our clients.

To give you an idea of Spacemaker’s capabilities for architectural programming, check out the screenshots below from a quick test that I ran with one of the demo sites that they offer.

Not only are the graphics clear and accurate, but the fact that Spacemaker operates via the cloud through a web portal opens up a lot of flexibility.

From here, Spacemaker allows you to run additional analyses on the architectural program models, such as:

• Wind Analysis
• View Analysis
• Solar Studies
• FAR Testing
• Facade Studies
• And more!

Within just a matter of minutes, you can create a variety of data-backed architectural programming models to include in your pre-design studies.

With AI, architects can customize designs based on the specific needs of their clients and make sure everything meets building codes and environmental regulations. Plus, it’s great for exploring new design ideas and experimenting with different materials and costs.

AI can save architects in the early stages of a project by using AI. No more crunching numbers and analyzing data for hours on end. Now you can spend more time being creative and coming up with awesome design concepts that will blow your clients’ minds!

So, if you’re an architect or designer, it’s time to embrace the power of AI. Trust me, it’ll make your job easier, more efficient, and way more fun.

Additional Resources for Architectural Programming

Want to learn more about architectural programming? Be sure to check out this post on 7 Essential Books for Architectural Programming and grab a few of these books for yourself!

Architectural Programming

7 Essential Books

The architectural program is the “playbook” for the design, so getting it right is crucial. If you’re looking for inspiration on the process for architectural programming, be sure to check out these 7 essential books!

Read More →

Jon Henning

Hi, I'm Jon. I write about emerging technology in architecture, engineering and design, and I want to help you push boundaries with the latest tech trends in the AEC industry.

Download the Free Guide: 100+ Realistic AI Architecture Prompts

Want to step up your AI Architecture Visualizations? Grab a copy of my FREE eGuide, which contains over 100 examples of AI Architecture prompts to use with MidJourney 5.

Keep Reading

Explore more →

AI In Architecture: The Good, The Bad, And The Unknown

How to See Original Stable Diffusion Training Images Using Datasette

How To Use AI In Architecture: 6 Phases Of The Design Process