Overview
Qdrant lets a cloud flow store embeddings, with data alongside each, in your own Qdrant and find the closest matches. The points stay in your cluster; Automize keeps nothing.
- Regions: Qdrant Cloud in any region, or a self-hosted Qdrant reachable from the internet over HTTPS.
- Plans: Cloud flows only. The bot has no Qdrant connector yet.
Setting up the connection
Auth type: API Key.
- In the Qdrant Cloud console, open the cluster and copy its URL, with the port (usually 6333).
- Create an API key for the cluster. A self-hosted Qdrant with no key set needs none.
- In Automize, open Settings → Connectors → Qdrant → Add connection, and paste the URL and the key.
Rate limits
Set by the cluster's size.
Data model
Cluster ─── Collection (one vector size) ─── Point
{"id": 1, "vector": [0.1, 0.2, …], "payload": {"source": "invoice.pdf"}}Available operations
5 operations available. Click any row to jump to its detail.
| Operation | Category | What it does |
|---|---|---|
| Create a collection | General | Make a collection for vectors of one size. |
| Delete points | General | Delete points by id, or the ones that match a filter. |
| List collections | General | List the collections in the cluster. |
| Find similar points | General | Find the points closest to a vector. |
| Upsert points | General | Add points to a collection, or replace the ones with the same id. |
Operations
Create a collection
Make a collection for vectors of one size.
| Parameter | Required | Type | Description |
|---|---|---|---|
Collection collection | Yes | text | The collection's name. |
Vector size size | Yes | integer | How many numbers each vector has; the embedding model decides this, e.g. 1536. |
Distance distance | Yes | select | How closeness is measured. Cosine suits most text embeddings. |
Tips
- The size must match the embedding model, e.g. 1536 for OpenAI's text-embedding-3-small.
Delete points
Delete points by id, or the ones that match a filter.
| Parameter | Required | Type | Description |
|---|---|---|---|
Collection collection | Yes | text | The collection's name. |
Point ids points | No | list | The ids to delete, e.g. [1, 2, 3]. |
Filter filter | No | object | Delete the points whose payload matches. Use this or Point ids. |
Wait until deleted wait | No | boolean | Answer only once the points are gone. |
Tips
- Give Point ids or a Filter.
List collections
List the collections in the cluster.
No input parameters.
Find similar points
Find the points closest to a vector.
| Parameter | Required | Type | Description |
|---|---|---|---|
Collection collection | Yes | text | The collection's name. |
Vector query | Yes | list | The embedding to match, e.g. [0.1, 0.2, …]. |
How many limit | No | integer | How many matches to bring back. |
Filter filter | No | object | Only points whose payload matches, e.g. {"must": [{"key": "source", "match": {"value": "invoice.pdf"}}]}. |
Include payload with_payload | No | boolean | Bring back each match's payload. |
Vector name using | No | text | For a collection with named vectors: which one to search. |
Tips
- The matches are in result.points, closest first, each with its score.
Upsert points
Add points to a collection, or replace the ones with the same id.
| Parameter | Required | Type | Description |
|---|---|---|---|
Collection collection | Yes | text | The collection's name. |
Points points | Yes | list | A list of {"id": 1, "vector": [0.1, 0.2, …], "payload": {"source": "invoice.pdf"}}. An id is a whole number or a UUID. |
Wait until stored wait | No | boolean | Answer only once the points are stored. |
FAQ
- Where do the vectors come from?
- From an embedding step, such as OpenAI's, earlier in the flow.