AI Semantic Search

What is Semantic Search?

Unlike traditional keyword search, Screvi's semantic search understands the meaning behind your query. It uses AI embeddings (mathematical representations of text meaning) to find highlights that are conceptually related to what you're looking for, even if the exact words don't appear in your highlights.

How to Use It

1. Go to the Search page (web app) or the Search tab (mobile app)
2. Type a natural-language question or topic
3. Results are ranked by semantic relevance
4. Click/tap any result to see the full highlight

Example Queries

• "What did I read about making better decisions?"
• "Quotes about resilience and overcoming challenges"
• "Ideas about the compound effect of small habits"
• "Stoic philosophy on dealing with adversity"
• "Business strategy for startups"
• "Parenting advice about setting boundaries"

The AI finds highlights that match the concept, not just the words. A search for "building good habits" might return a highlight about "the power of tiny daily improvements", even though none of the search words appear in the highlight.

Filtering Results

You can narrow search results with filters:

• Source type: Search only books, articles, tweets, etc.
• Tags: Limit to highlights with specific tags
• Favorites: Search only favorited highlights

How It Works (Technical)

Screvi generates an AI embedding (a vector in high-dimensional space) for each of your highlights. When you search, your query is also converted to an embedding, and Screvi finds highlights whose embeddings are closest to your query's embedding using cosine similarity. This is powered by AI text-embedding models.

Embeddings are generated automatically for all your highlights. New highlights get their embeddings within a short time of being imported.