AI-Lisa: how Enolisa's hyper-personalized pairing engine works
At Enolisa, we did not want a recommendation that sounded nice but stayed generic. We wanted recommendations that genuinely help you decide what to eat with a specific bottle, at a specific moment, for a specific person.
That is why we built AI-Lisa, our AI engine for hyper-personalized pairings. AI-Lisa is our decision system built on reasoning models, designed to turn sensory perception into actionable recommendations.
The problem we solve
Most traditional engines run on static rules: "if it is red, suggest meat." The issue is not only that this is simplistic, but that it ignores the most important factor: how that user is perceiving the wine right now.
In practice, two wines in the same category can behave very differently, and two users can have very different sensory experiences with the same bottle. If you do not incorporate signals such as body, intensity, finish, aromas, flavors, context, and cooking level, the result is no longer personalized; it becomes generic.
Our goal is precision with context.
Our approach: guided reasoning, not impulsive output
With AI-Lisa, we use Reasoning Models and a guided Chain of Thought (CoT) pattern to enforce a decision process before producing an answer.
We do not ask for one direct output. We ask for comparative evaluation. In practice, AI-Lisa thinks more like a sommelier than a static database.
Decision pipeline
The recommendation is not generated in one jump: it follows a three-phase flow that turns real tasting signals into a robust final decision.
01 Rich input signal
We do not stop at the bottle name. We feed AI-Lisa with sensory and contextual signals so the decision starts from real tasting data.
02 Alternative generation
AI-Lisa internally builds multiple clearly different pairing options, avoiding the obvious default answer.
03 Scoring-based selection
Then AI-Lisa compares those options with criteria defined by us (sensory coherence, structural balance, contextual fit, and cooking level, among others) and selects the best final alternative.
What this delivers for users
For people using Enolisa, this means:
- Hyper-personalized pairings with high relevance from the first recommendation.
- Useful explanations connected to their own sensory experience.
- Clearer, more confident food decisions in real consumption moments.
What this delivers for the business
For Enolisa, AI-Lisa helps us scale a pairing experience with premium depth and consistency:
- Clear differentiation versus generic recommenders.
- Stronger perceived trust in recommendations.
- A solid base to evolve product, retention, and long-term quality.
In conclusion, we go beyond generic recommendations to deliver gastronomic proposals with clear judgment, grounded reasoning, and coherence with each user’s sensory experience and real context.
An advanced system that is also available in freemium
One of the most important evolutions of this system is not only how sophisticated AI-Lisa has become, but how much more accessible that capability is today.
Enolisa now lets every user discover hyper-personalized pairings through a freemium model. On the free plan, each tasting can give access to a personalized pairing within a free quota over a 30-day window. Premium turns that into unlimited access and extends the experience with the full gastronomic layer: the complete explanation, the recipe when available, the structured analysis, and a living pairing library connected to your tastings.
That matters because this capability is no longer reserved for a narrow expert segment from day one. You can start using it for free, understand the value it brings to your own tastings, and go deeper later if you want the full accumulated experience.