What does someone mean when they ask for a digital twin? I brought Kelly Watt, Jason San Souci, Carlos Femmer, and Jeremy Pilgrim together to discuss the question from their different kinds of work. The terminology was part of the conversation, but the more useful connections concerned decisions, accuracy, adoption, and information that remains useful over time. These five takeaways give the discussion a practical structure.
Plan for use and change over time
Kelly separated the platform from the information and decisions around it. In his words, “The platform's not the digital twin.” He described a data model feeding a decision process, with boundaries that depend on the use case. That helps move the conversation from the name of the software to what someone needs to accomplish.
Carlos discussed a first capture as a baseline and a later capture as a way to understand change. The frequency should relate to what the asset and the decision require. The group did not need one update cycle for every kind of facility. The useful question is what another observation would help someone understand and how the information will be used again.
Match accuracy to the actual decision
Jason raised the relationship between accuracy, cost, and the consequence of a decision. An impressive specification needs to connect to the job it is meant to support. Jeremy illustrated the boundary with a carpet example: “Use this model to get a really, really good deal on your carpet, but just don't cut the carpet to match the model.” The same information can help with one task while being unsuitable for another.
The handoff needs to make that distinction understandable. Carlos discussed including survey information and metadata so a later user can see what was collected and how. A detailed-looking file may invite assumptions if the context is missing. Writing down the intended use and limitations helps the next person recognize what still needs to be checked.
Learn how the field team will use it
Jason described the risk of adding friction to an operator's day: “If it adds a step of work for the operator, it's gonna quietly die, politely.” That was his way of explaining why a promising program can stop being used without a dramatic objection. The buyer and the person doing the daily work may experience the system differently. Adoption needs attention to both.
Kelly described working through real workflows and helping people build a useful habit. Carlos emphasized a champion who actually wants to use the information. Conditions such as connectivity, a readable screen, or an understandable interface can affect what happens in practice. Those details become clearer when the team spends time with the people expected to make the change.
Keep project information useful afterward
The group discussed capture that is commissioned for one project and then becomes difficult for another team to find or use. Kelly's emphasis was on keeping context with the data: what it is, where it came from, and why it was collected. That gives a later user more than a file whose appearance suggests it might be helpful. The information needs a way to remain understandable beyond the original handoff.
Jeremy suggested building connections between systems rather than assuming every department should use one system for everything. Different teams have reasons for keeping different records. Shared references can help them recognize when those records concern the same asset or space. The practical question is which connection would help someone answer a recurring question in their work.
Keep clean data and experienced review involved
The AI discussion included examples the guests described from their own work. Those accounts are their experience, rather than independent performance tests of a tool or a promise of the same result elsewhere. Faster processing still leaves questions about the information supplied and what the output supports. The people responsible need a way to recognize a result that does not make sense.
That connects AI with the rest of the conversation rather than treating it as a separate answer. The capture, the context, and the review all contribute to whether someone can rely on the information. A new process should make those responsibilities clear enough for the team to discuss. The aim is a decision supported by understood information, rather than confidence created by the speed or polish of the output.
The four perspectives show why these conversations are useful across companies and project types. No one person has tested every approach, and another team's experience can help expose a question we have missed. Hear What Is a Digital Twin, Really? | Digital Twins Across the Built Environment on YouTube, Apple Podcasts, or Spotify.