Odoo's built-in search is functional — but it's database search. Full-text queries hit PostgreSQL, there's no typo tolerance, no instant-as-you-type results, and no relevance ranking. For product catalogs, knowledge bases, or any customer-facing search bar, that's a real limitation. This is what connector_typesense solves.

What Is Typesense?

Typesense is an open-source search engine built in C++, designed specifically for the use cases where Elasticsearch is overkill and Algolia is too expensive. The key characteristics that make it relevant for Odoo projects:

The core tradeoff with Typesense: it keeps its entire index in RAM, which is what makes it so fast — but it means your server needs enough memory to hold your searchable data. For most Odoo product catalogs (tens of thousands of products), this is not a problem. For datasets in the hundreds of millions, you'd look elsewhere.

How connector_typesense Works

The module sits inside the OCA search-engine framework — a generic connector layer that handles the binding, synchronization, and export logic between Odoo records and external search indexes. connector_typesense is the Typesense-specific backend for this framework.

The important design decision here: Odoo does not serve search queries directly. Records are indexed into Typesense asynchronously. The frontend (your website, portal, or external app) then queries Typesense directly, getting sub-50ms results without touching the Odoo database at all for search operations.

Installation and Setup

docker run -d \
  --name typesense \
  -p 8108:8108 \
  -v /path/to/data:/data \
  typesense/typesense:latest \
  --data-dir /data \
  --api-key=your-api-key \
  --enable-cors

# Verify it's running:
curl http://localhost:8108/health
# → {"ok":true}

./src/search-engine:
  remotes:
    origin: https://github.com/OCA/search-engine.git
  target: origin 18.0
  merges:
    - origin 18.0

# Modules needed:
# - connector_search_engine   (base framework)
# - connector_typesense       (Typesense backend)

Host:      localhost          # or your Typesense server IP
Port:      8108
Protocol:  http              # or https in production
API Key:   your-api-key      # set when starting Typesense

What Gets Indexed — A Practical Example

The module is designed to export any Odoo model to a Typesense collection. The most common use case is product search, but the framework is generic. Here's a simplified view of what a product index definition looks like:

# Defined via connector_search_engine binding + serializer
# Maps Odoo product.template fields → Typesense document schema

{
    "name": "odoo_products",
    "fields": [
        {"name": "name",         "type": "string"},
        {"name": "description",  "type": "string"},
        {"name": "categ_id",     "type": "string", "facet": True},
        {"name": "list_price",   "type": "float",  "facet": True},
        {"name": "active",       "type": "bool"},
        {"name": "website_url",  "type": "string"},
    ],
    "default_sorting_field": "list_price"
}

# Result: any search hits this index, not PostgreSQL
# Typo tolerance, faceted filters, and ranking — automatic

Once a record is bound, the synchronization layer watches for changes in Odoo and keeps the Typesense index up to date. Creates, writes, and deletes in Odoo propagate to the search index — so the data stays fresh without manual re-indexing.

Why This Matters for Odoo E-commerce

Odoo's website module uses PostgreSQL for product search. That works, but it has real limitations that become painful at scale:

• Typos produce no results — "iphone charger" won't find "iPhone charger"
• No relevance ranking — all text matches are treated equally
• No instant-as-you-type — each keystroke hits the database
• No faceted filtering without custom SQL — category, price range, attributes
• Performance degrades as the catalog grows

With connector_typesense, all of these are solved at the architecture level — the search engine handles it, and Odoo stays focused on business logic and order management.