OHMS

Search your digital assets using natural language queries. The AI understands context, dates, categories, and more.

Assets without embeddings:

All assets have embeddings generated. Semantic search is fully operational!

Ask naturally: "Show me unpaid invoices from last month" or "Find RedBalloon voucher"

Natural Language Search

Try:

How AI Search Works

Each asset is converted to a 768-dimensional vector using Ollama's nomic-embed-text model, capturing semantic meaning beyond keyword matching.

LLM parses your natural language query into structured filters (date ranges, categories, types) combined with semantic search terms.

PostgreSQL pgvector performs cosine similarity search on embeddings, combined with traditional metadata filters for optimal relevance.

Natural Language Query	What It Understands
`"Unpaid invoices from last 30 days"`	Date range filter + financial category + action filter
`"Find RedBalloon voucher"`	Entity extraction + semantic search
`"High priority emails from John"`	Priority filter + asset type + author filter
`"Photos from vacation in 2024"`	Asset type + date range + semantic context
`"Legal documents that need review"`	Category filter + action intent + asset type

System Requirements

Ollama Running Locally: Required for embedding generation

Make sure Ollama is running with the nomic-embed-text model:

ollama pull nomic-embed-text

PostgreSQL with pgvector: Already configured

Your database has pgvector extension enabled for efficient similarity search.