Canadian humanoid robot manufacturer Realbotix Corp. has completed delivery of its first commercially deployed robot equipped with Vinci, a proprietary AI vision and recognition system built directly into the robot’s eyes. The client, Ericsson, becomes the inaugural enterprise to operate a Vinci-enabled unit in a live environment.
The strategic weight of the Ericsson deployment lies as much in the client as in the product. Ericsson is a global telecommunications infrastructure company with operations across more than 180 countries. Its willingness to integrate an AI-equipped humanoid into an operational setting sends a signal to the broader enterprise market that the threshold for adoption is within reach. For procurement officers and technology strategists still in evaluation mode, that reference point matters considerably.
Operational value
Vinci’s defining characteristic is contextual continuity across interactions. By positioning cameras inside the robot’s eyes rather than on external mounts, Realbotix has built a system capable of maintaining natural, sustained eye contact while simultaneously running a suite of real-time perceptual tasks. The robot can identify an individual it has previously encountered, retrieve a record of that prior conversation and resume from where it left off — without any input from the user or a human operator.
In practice, this means the robot can recognise returning customers, track what those customers do and how they respond emotionally during an interaction, and generate structured reporting on engagement over time. The identity layer — knowing who is in front of the robot — underpins the rest of the capability stack. Behavioural tracking and emotional signal capture are only meaningful when they can be attributed to known individuals across multiple sessions, which is precisely what the facial recognition component enables.
For organisations running clinical trials, training programmes or customer engagement initiatives, the implications are significant. Longitudinal behavioural data that previously required dedicated research infrastructure, observation staff or self-reported feedback can now be collected passively as a byproduct of normal interaction. The robot does not interrupt the experience to solicit a rating or hand over a survey. It observes, recognises and records.
Andrew Kiguel, CEO, Realbotix: “Embedding cameras directly within the robot’s eyes allows the system to make and maintain eye contact for natural interaction with people and surroundings. Ericsson is leading the way in adopting this next generation of humanoid robotics capability.
Sector applicability
The use cases that Realbotix identifies cluster around three broad enterprise needs: customer engagement at scale, training and assessment, and research data collection. Across these categories, Vinci’s recognition and memory capabilities change the economics of what is possible. Personalised interaction no longer requires trained human staff to remember individual visitors. Assessment no longer requires post-hoc surveys. Research data no longer requires a separate observation protocol.
Competitive and investment context
Realbotix is listed on three exchanges — TSX Venture (XBOT), Frankfurt (76M0.F) and OTC markets (XBOTF) — an unusual public market profile for a company in a sector dominated by private-stage competitors. That visibility creates both scrutiny and credibility, and the Ericsson deployment arrives at a moment when investors are actively separating humanoid companies with real enterprise deployments from those still in pre-commercial development.
The Vinci patent is commercially meaningful beyond its technical novelty. In a market where hardware form factors are converging and large language model integrations are becoming table stakes, proprietary sensing architecture at the eye level represents a defensible position. Competitors can replicate the conversational layer; replicating a patented in-eye camera and recognition system with an established client deployment on record is considerably harder. The modular design — Vinci is compatible with any existing Realbotix robot — also extends the addressable base without requiring customers to replace hardware they have already purchased.
