Artificial intelligence is beginning to change how engineers design building systems. While AI cannot replace engineering judgment, it can dramatically reduce the time spent on repetitive modeling tasks, allowing designers to focus on reviewing decisions instead of manually creating them.
In this demonstration, we used Claude Code, Revit, and FireBIM to automate a significant portion of a fire alarm design. From placing smoke detectors and notification appliances to circuiting devices and generating voltage drop calculations, the workflow shows how AI can accelerate fire alarm design while keeping the engineer in complete control.
Best of all, everything demonstrated is available today and works with Revit 2022 through Revit 2027.
Connecting AI to Revit with FireBIM
The foundation of this workflow is FireBIM’s Model Context Protocol (MCP) server. The MCP server allows Claude Code to communicate directly with a live Revit model.
Instead of simply answering questions about fire alarm design, Claude can inspect the building, understand the geometry, place devices, organize circuits, and explain every decision it makes along the way.
This creates an AI-assisted design workflow rather than a simple chatbot experience.
Starting with a Design Plan
Rather than immediately placing hundreds of devices, Claude begins by analyzing the building model and asking the engineer a few key design questions.
These include:
- Applicable code basis, such as NFPA
- Notification appliance type
- Desired smoke detection coverage
Using those inputs, Claude reviews the Revit model and develops a proposed fire alarm design. It identifies rooms requiring smoke detection, heat detection, horn strobes, and manual pull stations before inserting a single family into the model.
The AI also proposes a phased implementation plan, beginning with a small pilot area before expanding to the remainder of the building. This allows the engineer to review the results early and provide feedback before the larger design is generated.
Automatically Placing Fire Alarm Devices in Revit
Once the design approach is approved, Claude begins inserting fire alarm devices directly into the Revit model.
During the pilot phase, it successfully placed smoke detectors, horn strobes, and other notification appliances while adapting to real-world modeling conditions. Architectural models are rarely perfect, so the AI encountered missing ceilings, curtain walls, and incomplete hosting surfaces.
Instead of stopping or producing an error, Claude evaluated the available geometry, made reasonable placement decisions, and documented every assumption it made.
After validating the pilot area, the AI completed the remainder of the first floor, placing devices throughout the building in just a few minutes of active engineering time.
AI That Explains Its Decisions
One of the biggest advantages of this workflow is transparency.
After each phase of the design, Claude provides a summary explaining exactly what it accomplished. It reports:
- Devices inserted
- Rooms completed
- Architectural modeling issues
- Design assumptions
- Items that require engineering review
For example, if a pull station cannot be hosted because the architectural model is incomplete, the AI identifies the issue and explains how it handled it. If ceiling heights are missing, it documents how those values were inferred.
Instead of functioning as a black box, the AI produces an audit trail that engineers can review before accepting the design.
Automatic Fire Alarm Circuiting
Once the devices are placed, Claude can organize the fire alarm circuits.
In this demonstration, the AI automatically created two SLC loops and three NAC circuits, assigning devices to each circuit and summarizing the completed work.
While Revit still requires wire graphics to be drawn manually, the circuit organization itself is completed automatically. This removes another repetitive task from the design process while allowing the engineer to verify the final layout.
Creating Fire Alarm Riser Diagrams
FireBIM extends Revit with tools specifically developed for fire alarm design.
After the devices and circuits were complete, FireBIM automatically generated a fire alarm riser diagram directly from the Revit model.
Because the riser is connected to the model, future design changes are reflected automatically. This eliminates the need to manually recreate riser diagrams every time the project evolves.
Performing Voltage Drop and Battery Calculations
Electrical calculations are another area where automation provides significant value.
Claude generated a list of installed devices so manufacturer load values could be imported into the Revit families. After receiving the load information, the AI populated the electrical characteristics throughout the project and even corrected unit differences when necessary.
With those values available, FireBIM generated:
- Point-by-point voltage drop calculations
- Battery sizing calculations
- Circuit loading information
- Panel load summaries
The voltage drop reports identify the available voltage at every device along each circuit, while the battery calculations determine the required standby capacity for the completed system.
During the review, the AI also identified one circuit that was approaching its voltage drop limit and suggested possible engineering solutions. Rather than simply performing calculations, it highlighted a condition that deserved additional review.
AI Supports Engineering Judgment
Although AI completed a large portion of the repetitive work, the engineer remained responsible for the final design.
The system intentionally identifies areas requiring professional judgment, including:
- Architectural modeling deficiencies
- Device placement compromises
- Voltage drop concerns
- Remaining design work
- Construction coordination items
This collaborative workflow allows AI to handle repetitive production tasks while engineers focus on design decisions that require experience and expertise.
Faster Fire Alarm Design with Revit and AI
The complete demonstration, including device placement, circuiting, riser generation, and electrical calculations, required about an hour of AI processing time.
Whether an experienced designer could complete the work faster depends on the complexity of the project. The real benefit is that Claude performs this work in the background while the engineer continues working on other aspects of the project.
Instead of spending hours placing families, organizing circuits, and documenting calculations, engineers can review AI-generated designs, make adjustments where necessary, and move on to higher-value engineering tasks.