Overview
SwiftSearch AI Search for Typesense is the only WordPress solution that enables native AI semantic and hybrid search capabilities in Typesense, adding advanced machine learning capabilities directly to your indexing cluster. Instead of restricting searches to exact keyword matches, AI search uses vector embeddings to map search terms conceptually, bringing up relevant products and content by context rather than spelling.
Semantic Matching
Find synonyms and related items automatically. Search queries like "pants" will match items labeled "trousers" or "slacks" without manual mapping.
Hybrid Relevance
Combine exact text matching and vector search. Adjust the alpha balance to optimize queries for your specific catalog type.
Auto-Embedding Pipelines
Schema generation is managed automatically. Typesense generates and saves vectors inside the database whenever products are synchronized.
Flexible Model Choice
Use built-in cluster models (GTE Small, MiniLM) to process vectors locally for free, or link external APIs like OpenAI for multilingual precision.
Addon Pricing & Licenses
Choose the billing plan that fits your search scale. All options license the full suite of SwiftSearch AI capabilities on a single site instance.
Monthly Plan
- Single Site License
- Native Local & OpenAI Vectors
- Hybrid Search Weight Tuning
- Auto-embedding Schemas
- Support & Updates (Active)
- Cancel or upgrade anytime
Annual Plan
- Single Site License
- Native Local & OpenAI Vectors
- Hybrid Search Weight Tuning
- Auto-embedding Schemas
- 1 Year Support & Updates
- Cancel or upgrade anytime
Lifetime Plan
- Single Site License
- Native Local & OpenAI Vectors
- Hybrid Search Weight Tuning
- Auto-embedding Schemas
- Lifetime Support & Updates
- One-time payment (No fees)
1. Installation
Download zip
Download the typesense-ai-search-for-swiftsearch.zip file delivered to your purchase email receipt.
Upload to WordPress
In your WordPress panel, go to Plugins > Add New > Upload Plugin, select the zip, and install.
Activate
Click Activate to initialize the AI Search settings panel under the main SwiftSearch settings menu.
2. Configuration & Re-indexing
Getting your AI search running requires a quick configuration and a data re-sync to generate the vector embeddings:
- Go to SwiftSearch > AI Search in your WordPress dashboard.
- Enter your license key in the Freemius connection screen to unlock the features.
- Toggle Enable Semantic (AI) Search to on.
- Select your AI Vector Model (built-in
ts/gte-smallis recommended for fast, local computation). - Choose the content fields to include in the AI Indexing Scope (e.g. titles and descriptions).
- Set your Hybrid Search Alpha Balance (0.0 to 1.0). Higher values weight towards semantic concept matches, while lower values prioritize exact keywords.
- Click Save Settings.
CRITICAL STEP: After saving, navigate to the Sync Dashboard of the core SwiftSearch plugin and run a full Re-index / Sync. This indexes your items under the new schema and generates the vector embeddings for your existing content.
3. Privacy & Connection Safety
We ensure that your data is handled securely throughout the vector generation process:
- Local Data Processing: When using built-in models like GTE Small or MiniLM, all vector math is handled entirely on your own Typesense cluster. Your product description text never leaves your dedicated server environment.
- API Key Encryption: If you select OpenAI embeddings, your API key is encrypted and stored locally in your WordPress database. The addon only uses it to sign search and indexing payloads.
4. Troubleshooting & FAQ
Do I need to pay for vector generation?
No, if you choose the built-in models (GTE Small or MiniLM), embedding generation is handled locally by Typesense for free. If you select OpenAI, you will need a standard OpenAI developer account with an active API key, and standard API fees will apply.
Why do my searches return no matches after enabling AI?
This is almost always because the products were indexed before enabling AI search, meaning their vector fields are currently blank. Go to the base plugin's Sync tab and click Re-index / Sync to compute and save the embeddings.
Can I use AI search for custom taxonomies or fields?
Yes. Any custom post types, custom tax terms, or text metadata configurations indexed by the base SwiftSearch plugin are fully supported. Simply ensure they are active under the Indexing settings.
What is a good starting Alpha Balance value?
We recommend starting with 0.8. This gives 80% weight to conceptual AI similarity matches and 20% weight to exact keyword matches. If you find exact matches are too weak, lower the value towards 0.5.