How AI Vision is Shaping The Future of Personalized Retail Experiences

Online shoppers often encounter recommendations based on their purchase history. In this 2023 Retail Refined interview, Melissa Gonzalez asks CyberLink’s Terry Schulenburg how AI vision and opt-in facial recognition could bring similar continuity to physical stores.

Schulenburg describes kiosks, loyalty programs and pickup experiences in which a customer chooses to enroll. The discussion explores the proposed convenience of recognition, the work retailers would need to do to integrate it, and the concerns that can make customers hesitate. The examples reflect his account of the technology and its potential at the time of the interview.

CyberLink’s path: from on-device video tools to opt-in recognition

Schulenburg introduces CyberLink as a company with long roots in video software, describing its tools as commonly bundled with new laptops. He explains that work in image quality and face modeling carried into consumer “try-on” experiences—particularly makeup—where realism depended on accurately mapping a face rather than simply overlaying color on a screen.

From those face models, he says CyberLink extended into facial recognition, emphasizing that recognition is intended for people who volunteer their face and provide consent. In his telling, CyberLink is not the brand a shopper opts into; instead, the company supplies a recognition “engine” that retailers and other organizations integrate into their own customer or employee workflows.

Matching online-style personalization in physical stores

A central thread is Schulenburg’s argument that e-commerce has trained shoppers to expect recognition and continuity: stores remember prior purchases, suggest complementary items, and surface relevant offers. He contrasts that with walking into a physical store where staff may not know the shopper, layouts differ, and the customer must search for deals or recommendations without the same history-driven prompts.

His proposed bridge is the in-store kiosk (or register-adjacent experience) that can identify a returning, opted-in customer and then tailor guidance—such as pointing to specific sections with items that fit past purchases. He notes that conventional sign-in (ID and password) can achieve a similar outcome, but he positions face-based recognition as lower friction, especially on shared public devices.

How Schulenburg describes customer enrollment

Asked how retailers onboard, Schulenburg describes a programmatic approach tied to an existing app and loyalty logic. In one example, he mentions an East Coast tobacco retailer with dozens of stores where age requirements and repeat purchasing make identity confirmation and convenience especially relevant to the customer journey.

In his example, a retailer invites a customer to enroll through an app and submit a selfie. Schulenburg describes that as a voluntary enrollment process that would allow recognition at pickup, reducing the need to present a barcode or repeat identification steps. His account explains the proposed customer journey; it does not establish the requirements for consent in every deployment.

He claims the resulting experience changes perceived wait time and effort: customers can be recognized quickly, move to pickup, and avoid standing in line. Just as important, he emphasizes that the customer initiated participation; the system is not, in his description, indiscriminately capturing faces and identifying people who never agreed to enroll.

Adoption and concerns about facial data

Schulenburg acknowledges the adoption headwind: he says facial recognition stories in the news skew negative, and he spends time reframing the difference between broad, ambient surveillance and limited, opt-in recognition. He argues that when recognition is tied to a volunteered identity—such as at a bank or within a loyalty workflow—people are more likely to see it as legitimate and useful.

To explain how comfort can change, he shares a workplace example: an enterprise site with thousands of employees and multiple turnstiles introduced a single face-enabled lane alongside traditional badge-based access. He describes initial hesitation followed by rapid migration to face entry as employees realized the convenience of not needing to carry, remember, or retrieve a physical credential.

Where AI vision goes next: demographics, “smart store” controls, and a five-year prediction

The conversation also moves from recognizing enrolled individuals to estimating aggregate demographic patterns. Schulenburg describes software estimates of age and gender presentation and suggests that retailers could use those estimates to understand changes in store traffic over the day. He proposes adjusting elements such as music or lighting in response. These are the speaker’s proposed applications, rather than findings from a measured retail trial.

Asked for a five-year outlook in 2023, Schulenburg predicts broader use of faces in transactions, travel, pickup and locker access. He stresses that participation should remain voluntary and selective. His retail argument rests on whether customers see enough value in the convenience and loyalty experience to choose to participate.