Architect's desk at golden hour with a white house model on a hand-drawn floor plan and a laptop showing the ChatGPT logo

2026

ChatGPT for Architects: 8 Practical Use Cases, Prompts, and the Maket Workflow (2026)

Learn how architects use ChatGPT in 2026: 8 real use cases, ready-to-use prompts, and a Maket workflow that turns AI briefs into buildable floor plans.

Patrick Murphy

CEO & Co-Founder

•

15 min read

ChatGPT has moved well past the hype phase, and the tool looks very different from the chatbot many architects first tried a few years ago. It now searches the web, reads uploaded images and files, generates and edits images, runs code, and can carry out multi-step tasks in agent mode. For practices of every size, that makes it a useful drafting accelerator, research assistant, and communication partner. It is still not a replacement for design judgment.

This guide covers how that works in practice: 8 concrete use cases, a ready-to-use prompt for each one, an honest look at what ChatGPT still can’t do, and a step-by-step workflow that connects ChatGPT ideation to editable floor plans in Maket. If you’re designing a home for yourself rather than for a client, our guide on how to design and plan your home with AI applies the same ideas for homeowners.

Why Architects Are Turning to ChatGPT in 2026

Adoption is real but still early. In a 2025 study by the American Institute of Architects, only 6% of architects said they regularly use AI for their jobs, and just 8% of firms had built AI into their practice, with another 20% working on it. The most common applications were chatbots, image generators, and grammar tools.

The reasons for interest are practical. Architecture involves a large volume of text-based work that is time-consuming but not uniquely creative: specification writing, client emails, RFI responses, meeting summaries, grant narratives, and accessibility checklists. ChatGPT handles this kind of work well when it gets a clear, structured prompt.

Cost pressure is another driver. Junior staff time is expensive. If a language model can produce a solid first draft of a client proposal in a few minutes, a senior architect can spend their time refining and approving rather than drafting from scratch.

The tool itself has also become far more capable. ChatGPT now runs on GPT-6 (rolled out in October 2026). You can upload a sketch or site photo and ask it to describe the design, list questions to check, or suggest program changes. With web search on, its answers include links to the pages it used. Agent mode can browse websites, work with uploaded files, and complete multi-step research tasks, and you can guide or interrupt it along the way.

What ChatGPT Can and Cannot Do for Architectural Practice

Before getting into use cases, a calibration check matters. ChatGPT is a general-purpose AI assistant. It is not a design tool, a licensed professional, or an authoritative code database.

What it does well:

  • Drafting, editing, and restructuring prose at high speed

  • Generating structured outlines, checklists, and tables

  • Summarizing long documents and uploaded files

  • Searching the web and linking to the sources it used

  • Analyzing data you upload, such as schedules or cost sheets, by running code

  • Producing concept images and mood boards

  • Explaining technical concepts in client-friendly language

  • Brainstorming options and alternatives when given clear constraints

What it still cannot reliably do:

  • Guarantee that a code citation matches the edition and local amendments your jurisdiction has adopted. Web search helps you check, but the answer is only as good as the page it found.

  • Produce editable, dimensionally reliable floor plans, sections, or construction documents. Its floor plan images can look convincing, but they are flat pictures with no walls to move and no measurements you can trust.

  • Read precise measurements off a drawing. OpenAI’s own image input guidance notes that the model struggles with tasks that need precise spatial localization.

  • Replace the licensed professional judgment of a registered architect

  • Provide stamped structural calculations or validated energy models. It can run rough numbers in code, but that is not engineering.

  • Know your site conditions, client history, or firm standards unless you give it that context

The professional responsibility point is critical. Any content ChatGPT produces for client deliverables, code research, or project documentation must be reviewed and verified by a qualified architect before it is used. AI-generated text is a first draft, not a final answer.

8 Practical ChatGPT Use Cases for Architects

1. Client Communication and Proposal Writing

Proposals, scope-of-work descriptions, and client update emails take real time to write. ChatGPT can produce a polished first draft of almost any client-facing document in seconds when you give it the relevant project context.

Example prompt:

“Write a 300-word project proposal introduction for a residential addition project. The client is adding a 400 sq ft primary bedroom suite to a 1960s ranch-style home in a suburban neighborhood. Tone: professional but warm. Include a brief statement about our collaborative design process.”

You then review, personalize, and adjust for the client relationship. The value is not the AI-generated final product. It is the drafting time you get back.

2. Design Documentation and Specification Drafting

Project specifications, basis-of-design narratives, and product section outlines follow recognizable structures that ChatGPT handles well. For standard residential or light commercial projects, you can ask it to generate a MasterFormat section outline, draft a basis-of-design narrative for a specific system, or write a scope description for a bid package. You can also attach manufacturer data sheets and ask it to draft from them directly.

Example prompt:

“Draft a basis-of-design narrative for an exterior wall assembly using fiber cement cladding over a ventilated rainscreen, rigid continuous insulation, and a fluid-applied air barrier. The project is a 3-story mixed-use building in Climate Zone 5. Use a technical but readable tone for a construction document set.”

Verify every technical claim against current product data sheets and your firm’s standard details before including the output in issued documents.

3. Building Code Research and RFI Responses

Code research is a natural fit. ChatGPT can explain the logic and structure of IBC, ADA, or NFPA requirements in plain language, help you frame a code research memo, and draft a response to a contractor RFI that covers the relevant section references. With web search on, it links to the pages it used, which makes checking faster.

Example prompt:

“Explain the IBC 2021 requirements for accessible parking spaces in a surface lot serving a new 50,000 sq ft medical office building. Summarize the ratio requirements, dimensional standards, and signage requirements in a memo format I can send to the owner. Cite the sections and sources you rely on.”

Critical caveat: always cross-reference ChatGPT’s summaries against the code edition your jurisdiction has actually adopted, including any local amendments.

4. Cost Narrative and Project Scheduling Summaries

Owners and lenders often need narrative descriptions of cost estimates and schedules that translate a contractor’s numbers into plain English. You can paste the figures or upload the estimate spreadsheet, and ChatGPT will draft the narrative. Ask it to check the arithmetic too.

Example prompt:

“I have a preliminary cost estimate for a 12,000 sq ft office renovation: $85/sq ft for demolition and rough construction, $45/sq ft for MEP upgrades, $30/sq ft for finishes, and $20/sq ft for FF&E. Write a 200-word cost narrative for the owner summarizing the estimate, the major cost drivers, and the contingency rationale.”

With a careful review pass, this gets you a usable owner summary in minutes rather than an afternoon of drafting.

5. Creative Ideation and Concept Development

Concept work is where your design judgment matters most, so treat ChatGPT as a divergent thinking partner rather than a designer. It can quickly generate alternative program arrangements, suggest precedents to research, or propose descriptive language for a concept that you then push further.

Example prompt:

“I am designing a small community library in a rural New Zealand town. The program is 4,000 sq ft including a reading room, children’s section, maker space, and community meeting room. Suggest five design concept directions, each described in two sentences, drawing on local vernacular architecture and the surrounding landscape.”

ChatGPT’s image generation can also turn the strongest directions into quick mood images for internal discussion. Treat them as atmosphere, not design. The goal is to get through the divergent phase faster so you reach real design work sooner.



6. Trend Research and Precedent Analysis

Staying current on material innovation, design trends, and emerging typologies takes time. ChatGPT can produce structured summaries of a trend, suggest precedent projects to research, or explain the principles behind a movement. With web search on, ask it to include links so you can check each project quickly.

Example prompt:

“Summarize the key principles of passive house design in residential architecture and list five widely published precedent projects that demonstrate high-performance envelope strategies, with a source link for each. I will use this as background for a client presentation.”

Use the output as a starting list, not a final bibliography. Open every source and confirm each project exists and matches the description before you cite it.

7. BIM Data Summarization

Turning exported BIM data (room schedules, door hardware sets, material takeoffs) into something an owner can read is slow work. Export your Revit or ArchiCAD schedules as a CSV or Excel file and upload it. ChatGPT can read the file, analyze it with code, and reorganize it for a specific audience such as the owner, contractor, or facilities manager.

Example prompt:

“I’ve attached a room-by-room finishes schedule exported from our BIM model. Reformat it as a plain-English finishes summary organized by floor, suitable for sharing with the owner’s interior design consultant, and flag any rooms with missing finish data.”

If the schedule lives in a spreadsheet, OpenAI’s ChatGPT for Excel and Google Sheets add-in can also build, update, and explain the workbook directly. Review changed cells before relying on them.

8. Accessibility and Compliance Checking

ADA compliance, Fair Housing Act requirements, and local accessibility ordinances create a documentation burden on every project type. ChatGPT can help you draft a preliminary accessibility checklist, explain the logic behind a specific requirement, or write a compliance narrative for a permit application.

Example prompt:

“Generate an ADA accessibility checklist for a new 8-unit multifamily residential building. Include requirements for accessible routes, parking, common areas, and unit features under the Fair Housing Amendments Act and ADA Title III. Format it as a checklist with section references.”

As with code research, verify every item against current federal and local requirements before the checklist is used in project documentation.

From ChatGPT Ideation to Floor Plan Execution: The Maket Workflow

ChatGPT is most useful at the front of a project, during brief writing, concept framing, and program definition. It can even draw a convincing floor plan image, but as we covered in Is there a ChatGPT for architecture?, editing that image precisely is another story. That is where Maket’s AI-powered floor plan software comes in.

Here is a three-step process that connects the two:

Step 1: Use ChatGPT to draft a design brief or spatial program

Ask ChatGPT for a structured program. For example: “Create a spatial program for a 2,400 sq ft single-family residence with 3 bedrooms, 2.5 bathrooms, an open-plan kitchen and living area, a home office, and a two-car garage. The site is a 60 ft x 120 ft infill lot. Note adjacency requirements and circulation priorities. End with a short summary of total area, room list, house shape, and the top three priorities.”

ChatGPT will return room sizes, adjacency notes, and design priorities. The closing summary is the part you’ll carry into the next step.

Step 2: Enter the program into Maket’s guided steps

As our FAQ explains, Maket doesn’t work from long pasted briefs. Its text input is limited to 300 characters, and the generator works best when you follow its guided steps: rooms, square footage, house shape, then priorities. Use ChatGPT’s summary to fill in each step. Maket then generates several editable layout options in minutes, with accurate dimensions, doors, and windows. This is the step ChatGPT can’t do on its own: turning a written program into geometry you can edit. If the client already has a plan, you can upload it as a PDF or image instead on a paid plan.

Step 3: Refine, visualize in 3D, and hand off

Open the closest option in the layout editor to move rooms, resize spaces, and adjust walls. Furnish it, view it in 3D, and create photorealistic renders to present several directions to the client. When a direction is approved, export it as DXF on a paid plan and continue in your CAD or BIM software. To see everything the platform covers, explore Maket’s full feature suite.

The Starwood floor plan in Maket, with rooms, doors, windows and furniture

The Starwood plan in Maket

TV-facing living-room render generated in Maket from the Starwood plan, showing two separate cream sofas and a round wooden coffee table.

Living room render generated in Maket from the Starwood plan

This workflow doesn’t remove the architect from the process. It compresses the time between a client brief and a presentable design option, leaving more time for the judgment-intensive work AI can’t do. For a deeper look at how AI floor plan tools work, our AI floor plan generator guide walks through the process step by step, and How to use Maket as an Architect shows the full workflow in the app.

Comparing AI Tools for Architects: ChatGPT vs. Specialized Platforms

Most practices won’t rely on a single AI tool. Here is how the main categories compare:

Tool type

Example

Strengths

Limitations

General AI assistant

ChatGPT (OpenAI)

Drafting, research with web search, file and data analysis, concept images

Floor plan images aren’t editable or dimensionally reliable; no construction documents

AI floor plan generator

Maket

Editable residential floor plans from guided inputs or an uploaded plan, 3D views, renders, DXF export

Residential projects only; design-stage plans, not construction documents

AI image generator

Midjourney, ChatGPT Images

Mood boards, facade studies, concept renders

Not technically accurate; no editable geometry

BIM authoring

Revit, ArchiCAD

Full construction documentation and coordination

Steep learning curve, high cost, slow for early-stage options

A practical split is ChatGPT (or a similar assistant) for language-heavy and research tasks, Maket for residential floor plan generation and early design iteration, and BIM software for construction documentation. For a full breakdown of how Maket compares to other AI floor plan tools, see our top-rated AI floor plan tools compared guide, or browse our head-to-head comparisons.

No single tool covers the whole workflow. The architect’s role is to orchestrate the stack and apply licensed professional judgment at each stage.

Limitations and Professional Responsibility When Using AI in Architecture

Using AI in professional practice carries real responsibility. A few key principles:

Accuracy is the architect’s responsibility. If ChatGPT drafts a code research memo that contains an error and that memo is issued to a client or authority having jurisdiction, the error belongs to the architect, not the AI. That applies to web-sourced answers too: a cited page can be outdated or cover a different code edition. Review everything.

Client disclosure. Clients may ask whether AI was used in project deliverables. Have a firm policy and be transparent when asked. In the AIA’s 2025 research, 90% of architectural professionals said they were concerned about issues including inaccurate outputs, authenticity, and transparency.

Data privacy. On personal ChatGPT plans, conversations can be used for model training unless you turn it off in Data Controls. ChatGPT Business, Enterprise, and Edu don’t use your inputs and outputs for training by default. Either way, don’t paste confidential client information, contract language, or personal data without a firm policy that covers it. Take extra care with agent mode and connected apps, which can access the email and files you give them, and which workspace owners can restrict.

Copyright and originality. AI-generated text and images sit in an evolving legal landscape. For anything that will be formally published or submitted as original work, consult your firm’s legal counsel on appropriate use.

AI as complement, not replacement. Tools like Maket and ChatGPT are meant to make architects more effective, not to replace the design judgment, professional responsibility, and licensed expertise a registered architect provides. The tools handle repeatable, text-heavy, pattern-matching tasks. The architect handles everything that requires contextual judgment, client trust, and legal accountability. To see where Maket fits in that picture, visit Maket for architects.

Frequently Asked Questions

Is ChatGPT accurate for building code questions in architecture?

ChatGPT explains the structure and logic of major codes like the IBC, ADA standards, and NFPA clearly, and with web search it links to the sources it used. It still can’t guarantee it has the edition and local amendments your jurisdiction adopted. Verify every code reference against the adopted code before it goes into a project document.

Can ChatGPT replace an architect or architectural software?

No. ChatGPT can draft text, analyze files, and generate images, including images that look like floor plans. It can’t produce editable floor plans you can measure, construction documents, or engineered calculations, and it can’t carry the legal responsibility of a licensed architect. Specialized tools like Maket handle editable floor plan generation, and BIM platforms like Revit handle construction documentation.

What is the best way for architects to prompt ChatGPT for design briefs?

Be structured and specific: include the project type, total area, room list with approximate sizes, site constraints, climate zone or location, and the audience for the output. Ask for a table or structured list rather than flowing prose, and ask ChatGPT to end with a short summary of total area, rooms, house shape, and top priorities. That summary maps directly onto Maket’s guided steps, which work better than pasting a long brief.

How does ChatGPT compare to tools like Maket for floor plan generation?

ChatGPT can write a detailed spatial program and generate an image that looks like a floor plan, but the image isn’t editable and its dimensions aren’t reliable. Maket takes your rooms, square footage, house shape, and priorities through guided steps, or starts from an uploaded plan on paid plans, and generates editable layouts you can refine, furnish, render in 3D, and export. The two are complementary: ChatGPT for ideation and briefing, Maket for spatial execution.

Is it ethical and professionally safe for architects to use AI-generated content?

Yes, with appropriate review and disclosure practices. AI output is a tool, not a final product. The obligation is to verify accuracy before using it in client deliverables, be transparent with clients when they ask about AI use, and protect confidential data by checking data controls or using a business workspace. In the AIA’s 2025 research, 90% of architectural professionals said they were concerned about issues such as inaccurate outputs, authenticity, and transparency, so a written firm policy is worth having.

What are the best AI tools for architects in 2026?

For most practices, the most useful stack combines a general-purpose assistant like ChatGPT for writing, research, and data tasks, an AI floor plan generator like Maket for early residential design and client presentations, and an image tool such as ChatGPT Images or Midjourney for concept visualization. BIM platforms remain essential for construction documentation. Our comparison of top-rated AI floor plan tools covers the leading options in detail.

How do I use ChatGPT to write a client proposal for an architectural project?

Give ChatGPT the project context: type, size, location, client profile, and your firm’s approach. Ask it to draft a 250 to 350 word proposal introduction that describes the opportunity, your design process, and the value you bring. Then adjust the tone to match your firm’s voice, add specific client details, and personalize any references to past work. The AI draft gets you past the blank page; the final proposal is yours.

Can ChatGPT analyze architectural drawings or blueprints?

Yes, within limits. You can upload photos, scans, and drawing files and ask it to describe what it sees, suggest program changes, flag possible circulation or accessibility issues, or draft a written description of a design. OpenAI notes the model struggles with tasks that need precise spatial localization, so treat any measurements it reads off a drawing as approximate and never as construction-grade.