Puma Deploys AI Concierge to Enhance In-store Experience

Puma debuted an AI "digital human" concierge named Dylan on April 13 at its Las Vegas flagship. The installation is a seven-foot-tall screen housing a lifelike avatar that can converse in 100+ languages, surface product recommendations, and act as a specialist in Puma running products. The project is led by Ivan Dashkov, Puma's head of emerging marketing tech, and was created in collaboration with Nvidia as part of a broader in-store technology push that includes foot scanners and racing simulators. Puma is positioning the system to augment staff rather than replace them; employees bring the selected items to customers after Dylan guides them.

The experience is a digital-human agent integrated with in-store systems and natural language capabilities. Key operational features highlighted by Puma and local news coverage include:

• •Multilingual dialogue, with automatic detection and handling of more than 100+ languages to serve diverse tourist traffic
• •Product specialization, where the agent can "double click" into running-shoe subcategories and advise based on running style and goals
• •Human-agent handoff, where a staff member completes physical retrieval and fitting, keeping the service tactile and human-driven

The deployment appears to use Nvidia tooling for real-time rendering and agent orchestration, and ties into Puma's product catalog and point-of-sale workflows. Puma describes the interface as a specialist assistant, not a transactional kiosk, which implies intent to combine retrieval of structured product metadata with conversational UX and potentially vectorized semantic search over product specs and sizing data.

This rollout sits at the intersection of retail AI, digital humans, and workforce augmentation. For practitioners, Puma's deployment illustrates three practical choices: prefer high-visibility experiential hardware (large-screen avatar) to attract engagement, invest in robust multilingual NLP to serve global foot traffic, and enforce a human-in-the-loop model to preserve service quality and trust. Puma is deploying AI as a customer-acquisition and engagement tool while it attempts a corporate turnaround; the company reported a €643.6 million loss for 2025 and has set aggressive brand and growth targets.

This is not a frontier-research release; rather, it is a real-world integration challenge. Key trade-offs include latency and TTS/voice quality for a public environment, safe and accurate product advice to avoid returns or brand damage, and privacy and data-retention governance for in-store voice interactions. Puma's framing of Dylan as a complement to staff is strategically important: it softens automation concerns and focuses on scaling expertise and language coverage.

Adoption and metrics. Track engagement rates, conversion lift, average transaction value when Dylan is involved, and error rates in product recommendations. Also watch for privacy disclosures and any limitations on recording or storing customer conversations. If Puma proves measurable uplift, expect other retailers to replicate a similar multimodal, human-in-the-loop concierge model.