Snaptrude's AI capabilities are organized into a set of agents, each responsible for a distinct step in the design workflow. These agents can be run as part of the full AI Stack (Design Mode) or as standalone agents in Program Mode via a dropdown menu.
Design Agent - Run complete generation of model along with site analysis, program creation & AI inspiration renders.
Research Agent - Conduct research and analysis for the project
Charts - Create charts based on program or analysis
Interpret - Convert tabular data to program in the 'Program Sheet'
Site Analysis - Define site constraints independently (v3.35.6)
Generate Program - Program Creation agent
Update program - Make changes/updates to the program
Assign Dimension - Space Sizing & Proportions agent
Assign Stories - Vertical Stacking agent
Learn how to implement these agents in workflows mentioned at the end: Recommended Workflows.
Design Agent/ AI Stack
Available in: Design Mode (AI Stack Step 1) · Program Mode (Standalone)
The Design Agent or AI Stack runs a sequence of up to 6 agents in order after the user has prompted project requirement. This can be triggered at project creation with the 'Kickstart with AI' option or it can triggered in a blank project from the program mode AI agents dropdown. It basically, creates a base project for you to begin working on. Each agent handles one layer of the design process:
Step | Agent | What it does? |
1 | Site Analysis | Site location, climate & other site context, FAR, setbacks, etc |
2 | Envelop Generation | Calculate total built-up area, footprint, building height, number of floors |
3 | Program Creation | Define spaces & program blocks with labels, areas & counts |
4 | Space Sizing & Proportions | Specify space dimensions or proportions |
5 | Vertical Stacking Plan | Story count, floor distribution of departments |
6 | Design Inspiration | Visualization of facade, materials, possible architectural style, etc |
Each step is to be approved by the user to ensure AI has the right data to move forward to the next step. This process takes about 10-12 mins depending on the size of the project to complete. Read in detail about Design Agent/AI Stack in this article.
Research Agent
Available in: Program Mode
An intelligent research companion for architectural research and program analysis. Instantly queries building codes, accessibility standards, and industry benchmarks, and generates charts and analyses directly within your program workflow. This can be triggered at project creation with the 'Kickstart with AI' option or it can triggered in a blank project from the program mode AI agents dropdown.
What you can do with it:
Code & Standards Research
Query building codes, ADA requirements, and zoning regulations by location and building type.
Get current regulations and standards with citation-backed responses.
Program Analysis
Run automated compliance checking and occupancy calculations using your program data.
Industry benchmarking against similar projects.
Risk assessment and efficiency analysis.
How to use it:
Open Program Mode and select Research from the mode dropdown. Or while creating a new project, select the 'Kickstart with AI' option. And in the prompt box, you will find a dropdown, select 'Research' option there.
Type your research query in the prompt box (e.g., "What are ADA requirements for hotel corridors in New York?"). Additionally, attach PDFs for better context for AI to generate results.
To analyze imported data, import your Excel file first, then prompt the Research agent to analyze it.
How the AI responds:
Responses may include both text explanations and data tables simultaneously (appearing only in the AI Sheet) .
The AI considers your entire program context while focusing on any selected elements.
Data can be pulled from imported Excel files, existing custom sheets, and program data simultaneously.
Note:
1. Average response time: ~2 minutes for standard queries. Complex prompts, large file uploads, or comprehensive research requests may take longer.
2. Research agent output cannot be passed directly to stack agents (e.g., "Change my dimensions according to this" will not work). Any standards from Research must be manually applied to the program or use the 'Update Program' agent.
Site Analysis Agent
Available in: Program Mode (Standalone)
Reads site context and produces a structured Site Analysis table that feeds into downstream calculations across the Stack panel, Site Analysis sheet, and Area Panel (FAR + total buildable area).
How to run it in Program Mode:
Open the mode dropdown in Program Mode and select Site Analysis.
Type your site constraints directly in the prompt (e.g., site area, FAR, zoning).
Send the prompt — the agent reads from your prompt and/or the canvas automatically.
Review the Site Analysis table. Re-run with a follow-up prompt to refine without creating duplicates.
Data sources used (in priority order):
Your prompt — always takes highest priority and overrides everything else
Any attached PDFs with the site information
Imported topo (lat–long metadata → zoning and FAR inference)
Site polygon drawn on canvas (for accurate site area)
Assumes values if required data is missing
Key behaviors:
Re-running updates the existing analysis — it does not create a new duplicate table.
Geometry edits (rotate, cut/fill, redraw) do not invalidate site identity.
Site Analysis never modifies canvas geometry.
Other agents can reference site constraints as shared stack context.
You can define site before or after generating a program — both orders work.
Known Limitations:
The Select-Site modal is not available in Program Mode.
If no lat–long metadata and no zoning constraints exist, zoning and FAR will be assumed.
Once you explicitly define a value (e.g., site area via prompt), that value takes precedence over any newly imported parcels or drawn site polygons.
In Design Mode, the AI Stack can read topo metadata, but the Envelope does not yet read site polygon meshes.
Drawing multiple overlapping site polygons creating an excessively large site area may generate a very high Total Buildable Area and impact system performance.
Generate Program Agent
Available in: Program Mode (Standalone)
Creates an initial space program from scratch based on your brief — building type, GFA, and program mix.
How to use it:
Open the mode dropdown in Program Mode and select Generate Program.
Prompt with your building type and total GFA (e.g., "Design a program for hotel and retail mixed use. GFA is 120,000sf").
Review the generated program in the Program table.
If changes are needed before dimensioning, switch to Update Program to refine without regenerating.
Note:
1. Running Generate Program will overwrite an existing, refined program. Use Update Program instead if you want to make targeted edits.
2. Changes made by Generate Program do not appear in Program Mode immediately — visit the Design Canvas tab first to trigger the update.
Assign Dimension Agent
Available in: Program Mode (Standalone)
Batch-applies dimensions to all spaces in your program.
How to use it:
After your program is ready, open the dropdown and select Assign Dimension.
Send a general prompt (e.g., "Assign appropriate dimensions") — the agent applies dimensions in bulk across all spaces.
Review the results. For any specific dimension that needs a precise value, switch to Update Program (e.g., "Make all hotel rooms 15' deep").
Note: Assign Dimension cannot make surgical, space-specific edits. It applies dimensions broadly. For precise control, use Update Program after running this agent.
Update Program
Available in: Program Mode (Standalone)
To make any changes to the generated program, use the Update Program agent. Select the agent from the dropdown, then prompt the changes required. For instance, if you wish to change the space labels, areas or count, mention what needs to be changed, and AI will then update the program accordingly.
Assign Stories Agent
Available in: Program Mode (Standalone)
Distributes your program spaces vertically across floors.
How to use it:
Once your program and dimensions are finalized, open the dropdown and select Assign Stories.
Prompt with your storey count and any floor-specific placement preferences (e.g., "Stack spaces across 10 stories" or "Move private offices to levels 3 and 4").
Review the stacking plan and refine with follow-up prompts as needed.
Chart Agent
Available in: Program Mode (Standalone)
Create charts from your program or analysis using this agent. Simply select the agent from the dropdown in the Program Mode and then enter your requirement as a prompt. AI will then create charts in a new sheet which you can then edit/update as required.
Interpret Agent
Interpret converts data from your custom sheets directly into a structured Program Data Sheet. Select the relevant cells, prompt the agent, and it automatically extracts space names, areas, counts, departments, and other metadata into usable program data. Learn in detail how to work with this agent here: Interpret AI Agent.
Recommended Workflows
Starting from Scratch
Site Analysis — Define site constraints via prompt
Generate Program — Create initial program mix with GFA
Update Program — Refine spaces, quantities, or mix without regenerating
Assign Dimension — Batch-apply dimensions across all spaces
Update Program — Fix any specific dimension values precisely
Assign Storeys — Stack the program vertically
Already Have Program Data
Update Program — Make targeted adjustments (quantities, mix, individual spaces)
Assign Dimension — Apply batch dimensioning
Update Program — Apply specific dimension edits (e.g., "Elevator shaft must be 10' wide")
Assign Storeys — Stack and refine floor assignments
Research → Design Workflow
Research — Query building codes, ADA standards, or benchmarks for your building type and location
Interpret — Use the Interpret Workflow to translate research findings into spaces and departments
Update Program — Manually apply any specific standards from Research that need precise values
Assign Dimension / Assign Stories — Continue with the standard design workflow
Important Limitations to Know:
Research agent output is not readable by stack agents. Any data from Research must be manually re-entered or Update Program.
The Design agent cannot read program data — it is unchanged and operates independently.
User-provided values always override detected or assumed values in Site Analysis.
Selected data in your sheets receives higher priority in AI analysis — select relevant rows before prompting for analysis or charts.