AI Classification

Every transaction in Bliss is classified by a 4-tier pipeline that learns from your corrections over time.

The 4-tier waterfall

Each transaction flows through these tiers until one succeeds:

Tier 1: Exact Match

An O(1) in-memory cache lookup by description. If you’ve categorized “NETFLIX” before, the next occurrence is matched instantly with confidence 1.0.

Tier 2: Vector Similarity (Tenant)

If no exact match, Bliss computes a vector embedding (Gemini, 768 dimensions) and queries pgvector for cosine similarity against your previous classifications. Threshold: reviewThreshold (default 0.70).

Tier 3: Vector Similarity (Global)

Same as Tier 2 but searches across all tenants’ embeddings. Results are discounted by 0.92x since cross-tenant matches are less reliable.

Tier 4: LLM

If no vector match, Gemini 3 Flash (gemini-3-flash-preview) classifies the transaction with temperature 0.1. Confidence is hard-capped at 0.85 — LLM results always go through review.

Enabling AI classification

Tier 1 works out of the box. Tiers 2-4 require a Gemini API key (Tier 2 needs it to generate embeddings, Tiers 3-4 use it for cross-tenant search and LLM classification):

GEMINI_API_KEY=your_api_key

Without it, unmatched transactions remain unclassified for manual review.

Training the model

The most important thing you can do is review your first ~100 transactions carefully. Every correction you make:

1. Updates the exact-match cache immediately (Tier 1)
2. Generates a vector embedding asynchronously (Tier 2)

After this initial training period, most transactions will match at Tier 1 or 2 with high confidence.

Tuning thresholds

Two tenant-level thresholds control the pipeline behavior:

Adjust these in Settings based on your comfort level. Lower autoPromoteThreshold = less manual review but more risk of misclassification.

Expense tracking

Once classified, your transactions power the expense tracking dashboard with category breakdowns, monthly trends, and spending analysis.

Next steps

• Importing transactions — the review flow where you train the AI
• Explore the AI Classification spec for full technical details