Pain Point Observation：
Choosing from a menu can become a public performance
When a line is moving and a barista is waiting, customers often fall back on a familiar drink—even when they want to try something new.
Dense menu boards, unfamiliar customization language, uncertainty about extra charges, and the pressure to decide quickly make discovery harder at exactly the moment it should feel enjoyable.
PRODUCT OPPERTUNITY
move discovery into a private, low-pressure moment before ordering, while preserving each café’s menu, brand, and operational control.
Second Cup was selected only as a reference implementation because its public menu provides a realistic product catalogue for prototyping. The underlying system is designed to work with different café brands and independent coffee shops. This independent portfolio concept was not commissioned by, endorsed by, or created in affiliation with Second Cup.
BUSINESS MODEL
One recommendation layer, configured by each café
→
Cafe Knows
-
Recommendation engine
-
Partner console + QR hosting
Partner cafés
-
Store rules + brand control
-
Subscription hypothesis
→
Café customers
-
No app required
-
Recommendation before ordering
Cafe Knows operates between partner cafés and their customers.
Cafés configure their own menu data, store rules, branding, and location experience.
Customers then receive explainable recommendations based only on that café’s available offerings.
A real menu, used to test a reusable system
Second Cup’s public menu was used only as a realistic reference implementation. The underlying system is designed to work with different café brands and independent coffee shops. Official menu facts remain separate from inferred taste and pairing tags, while prices, availability, ingredients, and allergens require validation from each partner café.
PRODUCT LOGIC
01
Validate Menu
Map menu items, customizations, cautions, and store-specific rules.
02
Capture Intent
Collect visit context, taste, and hard constraints in one short session.
03
Filter + score
Remove conflicts, then weight menu fit and data confidence.
04
Diversify top 3
Return a safe choice, a small twist, and a bolder alternative—not three near-duplicates.
MVP scoring model
Fit = intent + menu attributes + store rules + confidence
Hard constraints are filters, not weighted preferences. A café must validate uncertain menu data before publishing.
What can learn later
With consent and sufficient pilot data, ranking can incorporate recommendation feedback and store-level outcomes.
Not in this prototype：
Payment · inventory integration · account history · medical or nutrition advice
CUSTOMER EXPERIENCE
Guided turns explicit preferences into recommendations. Vibe begins with a feeling, then confirms caffeine, milk, dietary, allergy, and availability constraints before ranking real menu items.
Guided Path
Vibe Path
Transparent by design
1 / Capture & filter
Capture visit context, taste preferences, or mood tags. Apply caffeine, temperature, milk, dietary, allergy, and store-availability boundaries before ranking.
2 / Score & diversify
Score what remains across taste, texture, context, mood, and preparation fit—then return a safe choice, a small twist, and a bolder alternative.
3 / Explain & translate
Explain why each result fits, lower confidence when menu data is incomplete, and translate customization into a barista-readable order card.