How the Digital Twin Powers Saffron AI

In modern building management, complexity often stands as a barrier between users and their facilities' full potential. Traditional methods of navigating building management systems (BMS) and the messy data they produce can be like a maze without a map, leaving users frustrated and unable to extract meaningful insights or actions. In commercial buildings, this changes forever at the intersection of the digital twin and conversational Artificial Intelligence (AI). 

Earlier this year, 75F launched Saffron AI: a conversational AI for commercial buildings. Using tagged data from a customer’s building, Saffron answers user questions, analyzes their space, and even acts on simple tasks like changing the temperature or building schedule. All of this comes packaged in a user-friendly chat interface so anyone — whether they have specialized BMS knowledge or not — can interact with their building in a meaningful way. 

To some readers, this may sound like the chatbot you use to cancel a subscription or initiate a product return. Yet, Saffron AI is so much more. As a closed ecosystem, Saffron AI is a personalized and trusted virtual assistant capable of far more than simply answering questions based on Internet searches. To understand the power of Saffron AI, we must first understand the vital role of the digital twin in commercial building management. 

The Digital Twin in Buildings

A digital twin is a virtual replica of a physical asset, system, or process. It mirrors the real-world entity in real time, enabling comprehensive analysis, prediction, and optimization. It’s important to think of the digital twin as a living entity — unlike a 3D model that statically replicates a physical asset, the digital twin represents the data flowing from systems every minute of every day. 

In most commercial building management scenarios, the data that comes from building equipment and BMS devices is hard to digitize in a uniform digital twin. This is because most data is tagged according to the equipment manufacturer or systems integrator who designed the system. Even when using Project Haystack to define equips, not every person uses the same combination of equips to define entire pieces of equipment, which can lead to data silos between systems in the same building. What a mess. 

75F solved this problem with a tool we created called Hayloft, an application focused on developing a standardized approach for organizing, tagging, and accessing data from various equipment within the built environment. Using Hayloft as a common repository, we define the digital twin of certain types of equipment only once and then reuse it across all applications.  

Saffron AI can do everything it does thanks to our spotless digital twin blueprint.  

Unlike conventional AI chatbots, Saffron AI operates within a closed ecosystem, leveraging real-time building data consistently tagged according to Project Haystack and Hayloft definitions, industry standards for HVAC operation like ASHRAE Guideline 36, and insights from the 75F support portal. This unique approach ensures that user interactions with Saffron AI are not only reliable but also highly tailored to the specific needs of their building, all without irrelevant noise from the Internet. 

What Can Saffron Achieve With Clean Data?

Now that we understand the concept of a digital twin and how it makes Saffron AI more reliable than an Internet-based chatbot, let’s dive into some of Saffron’s core capabilities and differentiators: 

• Performs real-time inference of user intents to explore, analyze, and act on commercial building data.

• Conversational style driven by multiple memory-backed autonomous agents that can reason with themselves, learn from experience, and adjust for and learn from past user behaviors and feedback.

• Because users can ask a huge variety of questions, Saffron's visualizations are driven by dynamic Haystack-based aggregation operations rather than being hardcoded.

• Computer-vision models allow Saffron to extract features from user-uploaded sketches, meaning users can ask Saffron to look at a picture of a dashboard they'd like to build and Saffron will build it for them.

• Saffron's recommender system integrates the rich ecosystem of existing 75F visualizations to suggest the most relevant and most-used widgets for users' queries.

Saffron AI represents a paradigm shift in building management, transforming raw data that is historically untapped into actionable insights and empowering users with newfound clarity and control over their facilities. By bridging the gap between complexity and accessibility, Saffron AI is ushering in a new era of commercial building management — all with trust and accuracy as the foundation.