Legacy discovery
What do other people think about this wine?
- Large database and crowd ratings
- Helpful after a bottle has already been found
- Requires the user to interpret quality and fit
- Can create decision paralysis
Not a wine database. A trusted decision engine.
The digital sommelier for people who want the right bottle, not more ratings.
Glass Half Full brings together real sommelier judgment, curated wine data, and AI reasoning to make wine feel clear, personal, and approachable.
- 60 starter wines in the seed dataset
- 5 style families across red, white, rose, sparkling, and fortified
- 0 pay-to-play placements or crowd-score dependency
Positioning
Move the conversation from popularity to personal fit.
The docs consistently frame Glass Half Full as a proactive advisor rather than a reactive lookup tool. That difference drives the entire experience.
Glass Half Full
What wine will you actually enjoy tonight?
- Sommelier-curated source data
- Guided before or during the buying decision
- Recommendations shaped by taste, budget, and occasion
- Built to reduce overwhelm and increase confidence
Approach
Human-first expertise, with AI used to scale judgment rather than replace it.
The product vision is anchored in trust: only recommend wines that would be confidently recommended in person.
Start with the decision
Ask for a few real-world inputs such as flavor preference, budget, food pairing, or occasion instead of forcing users to browse endlessly.
Reason over a curated set
AI matches users against a sommelier-shaped dataset instead of trying to infer quality from public sentiment and noise.
Deliver honest guidance
The system can suggest better-value alternatives, acknowledge trade-offs, and keep the tone simple enough for beginners.
Principles
Trust is part of the interface.
These principles come directly from the sommelier docs and shape both product behavior and business model.
Curated Collection
A starter wine base designed for recommendation quality, not content volume.
The seed dataset spans beginner-friendly everyday bottles, celebratory picks, and distinct recommendation lanes across styles and budgets.
Every entry is intended for recommendation logic, not filler inventory.
Records include pairing and occasion signals such as steak, patio, gift, and brunch.
The dataset notes future fields for reviews, partner inventory, and internal scoring.
Execution
Ship a focused MVP, validate quickly, then grow the moat.
The roadmap in the docs is disciplined: start with curation and core recommendation logic, then learn from real-world usage before expanding.
Planned stack
Lean architecture for a fast first launch.
- FrontendReact or Flutter
- BackendFirebase Functions
- DatabaseFirestore
- AIOpenAI API
Vision
Make expert wine guidance feel available to anyone, not just people standing in front of a great wine shop.
Glass Half Full has a clear strategic lane: trusted recommendations, simple decision support, and a stronger long-term taste relationship than crowd-scored platforms can offer.