BWB Episode 101 - The Boys are Back

After a hiatus, Brewing with BIM is officially back! Hosts Joe Whitney and Dave Campbell recently sat down to catch up on everything happening at the intersection of Building Information Modeling (BIM) and Artificial Intelligence.

While AI has been a major buzzword at Autodesk University and across the AEC sector for the past few years, Dave and Joe focused the conversation on practical application over hype.

Here are a few highlight topics from the episode:

The Power of Model Context Protocols (MCPs)

While out-of-the-box AI assistants are gaining traction, custom MCPs allow teams to bridge LLMs (like Claude) directly with design software, including Revit, Civil 3D, and Navisworks. By creating custom connections, users can query models for warnings, track revisions, and automate tedious modeling steps that native software still struggles with.

Avoiding the Token Trap: The Need for an Orchestrator

One major hurdle with AI integration is the cost and token load of querying large, data-heavy models. If you send massive prompts continuously across multiple models, costs spiral quickly. To combat this, the duo discussed using an orchestrator (or conductor) strategy—a lightweight layer that routes specific parts of a query to the precise MCP needed, aggregating filtered context back to the AI rather than dumping entire project files at once.

Enhancing Efficiency, Not Replacing People

Contrary to the fear that AI will replace jobs, Joe and Dave emphasize that AI’s true value lies in making specialized professionals more effective. By automating repetitious tasks—such as converting .ITM files into generic Revit families, auto-dimensioning, or exporting schedule CSVs to create baseline benchmarks—designers and engineers can focus on higher-level problem-solving.

The Importance of Prompt Libraries

As AI adoption grows across organizations, user variation becomes a challenge. The same request worded differently by two team members can yield vastly different results. Building a structured internal prompt library helps standardize outputs and ensures team-wide accuracy when querying models or generating project documentation.

🎧 Listen to the full episode now on Spotify or your favorite podcast platform!