Developer Integrations

Opensolr inside your code

Five packages put an Opensolr Index behind the frameworks people use to build search and AI apps. Install one, give it your account email and API key, and the meaning vectors, the search and the answers all run on Opensolr servers. There is no embedding model to run and no vector database to host.

The packages

What every package does

• Words and meaning, ranked together: hybrid search, with a balance you can tune. A keyword-only mode skips the meaning step and uses none of your AI allowance.
• Writes through Data Ingestion: your documents are queued, then enriched like a crawled page. See Push your data.
• Grounded answers: one call finds the best documents and writes an answer from them.
• Search with a photo: the words, text and barcodes in a picture become the query.
• Search operators: "phrases", +word and -word work the same everywhere. See Search operators.

The Python packages and the MCP server also read local files: a PDF page by page, and scans and pictures through OCR.

What your index needs

• Data Ingestion, search by meaning and PDF pages need an index on a server with site search and the Opensolr schema. Keyword-only search works on any index with title, description and text fields. See What your index needs.
• Each package checks an index once, never on every search, and says exactly what it cannot do and how to fix it: Check and install the schema.
• Meaning vectors, answers and pictures use your monthly AI allowance: API Quota.

Try it, or use a CMS

• Every package works with the public demo account, without signing up.
• Running Drupal or WordPress? Use the free module or plugin instead: Drupal, WordPress.
• A chat bot for your site, hosted or on your own server: Three ways to have your chat bot.
• Plain HTTPS from any language: API Reference.