Organizations adopting telepresence robots for remote collaboration - in offices, hospitals, classrooms - tend to treat communication quality as a hardware problem: better cameras, smoother mobility, lower latency. But teams don't just use a telepresence robot once; they build a working relationship with each other, and that relationship has a history. Brands and institutions lack clear data on whether how a team first met shapes how well they collaborate once a robot enters the picture.
We ran a within-subject, counterbalanced study in which 32 pairs (64 participants) each completed two comparably difficult escape-room tasks - one collaborating in-person, one through an Ohmni telepresence robot - with one partner acting as teleoperator and the other as the local interlocutor. Neither partner had all the information needed to solve the puzzles alone, forcing continuous verbal coordination in both conditions.
Half the pairs met in-person first and then moved to robot-mediated collaboration; the other half met through the robot first and then in-person. This let us separate two questions: what changes when a robot enters a collaboration, and does it matter which way the team first met each other?
We analyzed nearly 7,000 spoken utterances for verbal volume, turn-taking rhythm, vocabulary diversity, and collective ("we") versus individual ("I") language, and paired the linguistic data with post-task debrief interviews.
Our research showed that robot mediation doesn't make collaboration worse - it reconfigures how people do it, and the shape of that reconfiguration depends on how the team first met:
Robot Mediation Doesn't Silence Teams - It Makes Them Talk More: Participants produced roughly 124 more words per speaker when collaborating through the robot than in person. Losing the ability to point, gesture, and share a physical view didn't reduce collaboration - it pushed it into speech.
Teams Shift From "We" to "I": The clearest change was in how partners framed their contributions. In person, collective language ("we," "us") accounted for roughly 40% of first-person pronoun use; through the robot, that dropped to roughly 13%. This isn't a breakdown in teamwork - it reflects partners needing to be more explicit about who has which information and who takes the next action when they can't rely on shared physical cues.
Conversational Rhythm Tightens Under Robot Mediation: The gap between one person finishing and the next person starting to speak shrank by roughly 3.5 seconds under robot mediation, indicating a faster, more clipped back-and-forth - even though the total number of conversational turns didn't reliably increase.
How You First Meet Shapes What Happens Next: Teams that met in person first showed a real disruption to their rhythm when they switched to the robot - shorter turns, a choppier exchange. Teams that met through the robot first carried that same clipped, explicit style with them even after moving to in-person collaboration. The first encounter set a pattern that persisted.
People Still See the Human Behind the Robot: In debriefs, interlocutors consistently recognized the teleoperator as a real person controlling the system, even while describing the interaction as less warm and less engaging than in-person collaboration. Teleoperators, meanwhile, struggled most with concrete limitations - showing materials, referencing shared objects - rather than a sense of not being recognized.
The biggest lever for successful telepresence collaboration isn't a better robot - it's a better-structured first encounter. Teams don't just adapt to a robot in the moment; they carry the communication habits from their very first meeting into every collaboration that follows. For remote work, telemedicine, and education, this reframes the problem: instead of engineering away every limitation of the robot, organizations can deploy a simple, low-cost intervention - starting relationships in person - to build the relational capital that helps teams stay coordinated and connected once the robot takes over.