Broadedge Wearlink Co., Ltd. / PR TIMES press material (no separate photographer credit stated in the release)
WINE BISTRO Sommia opened in Ebisu, Tokyo, on August 6, 2026, as the restaurant half of an AI-driven wine concept built by Broadedge Wearlink. Instead of leaving pairing to guesswork or traditional sommelier language, the bistro uses Sommia, an AI trained with input from 30 leading sommeliers, to recommend wines dish by dish and explain the logic behind each match. The concept is connected to the official Sommia platform: https://sommia.jp/bistro.
The Concept
WINE BISTRO Sommia turns wine pairing into an interactive part of the meal rather than a one-off recommendation. Guests order from a bistro menu, then use Sommia to understand which wine fits a dish and why. The system is built around knowledge gathered from 30 professional sommeliers, but the experience is intentionally conversational rather than academic: it translates pairing logic into plain language and lets users rate what they drink so the AI can build a clearer picture of their taste. The bistro keeps 17 wines available by the glass, spanning Champagne, white, orange, rosé and red, with glasses starting in the ¥800 range. A separate AI WINE SHOP Sommia operates one floor above with more than 1,000 bottles from over 20 countries. Guests can buy a bottle there and bring it into the bistro under a BYO arrangement, linking retail, recommendation and dining in one system.
What Is Actually New
The genuinely new part is not simply putting a chatbot in a restaurant. Sommia connects three functions that are usually separate: personalised wine discovery, restaurant pairing and retail purchase. The AI does not only suggest a bottle; it also explains the pairing, records feedback and can continue helping the same user outside the restaurant through the Sommia app. That means the recommendation layer can follow the customer from a meal to future purchases instead of disappearing when the bill arrives. The adjoining shop strengthens that loop: its 1,000-plus wines are organised into taste-based categories, while the bistro provides a real-world tasting environment in which preferences can be tested. AI-driven recommendations already exist in retail, but using the restaurant itself as a live learning environment for a persistent personal wine profile makes this concept substantially more interesting than a standard digital wine list.
Why People Want This
Wine remains one of the restaurant categories where choice can feel disproportionately difficult. Grape variety, region, vintage, style and food pairing create a knowledge barrier that can make people default to familiar bottles or simply ask staff to decide. Sommia attacks that friction without removing discovery. A guest can ask questions privately, understand why a pairing works and gradually build a personal taste profile rather than memorising wine terminology. The appeal is therefore less about replacing sommeliers than lowering the social and knowledge barrier around them. It also adds a game-like feedback loop: every tasting can make future recommendations more personal. For curious but non-expert drinkers, that combination of guidance, explanation and progressive personalisation could make wine selection feel more approachable while still preserving the sense of exploration that makes restaurant wine interesting.
WBC Verdict
Sommia is compelling because the technology has a clear job: make a notoriously intimidating category easier to explore without flattening it into a generic recommendation engine. The strongest idea is the closed loop between restaurant, AI profile and 1,000-bottle shop. The open question is whether guests will actually keep rating wines and using the system often enough for personalisation to become meaningfully better. If they do, the restaurant is not just serving pairings; it is building a portable taste memory that could influence what customers order and buy long after dinner.