> ## Documentation Index
> Fetch the complete documentation index at: https://docs.raily.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Add a Vector Store

> Connect a vector database to Raily so your collections become searchable

Connect a vector database (Qdrant, Pinecone, and others) to Raily. Once connected, Raily can search its collections with access control and usage tracking on top.

The steps are similar for every provider. Pick your provider in the tabs below for its connection fields and steps.

## What you can do

Once a vector store is connected, Raily searches it with:

* **Semantic search** - find content by meaning, not just keywords.
* **Metadata filtering** - narrow results by date, author, category, or any field in your [Data Catalog](https://docs.raily.ai/integrations/data-catalog.md).
* **Hybrid search** - combine semantic similarity with keyword matching.
* **RAG** - feed retrieved passages to an LLM so its answers are grounded in your content.

Access control and usage analytics apply to every search, whichever provider you connect.

## Before you start

* Admin access to the tenant.
* The connection details for your vector store (URL, API key, or connection string).
* At least one collection (or index/table) that already holds embeddings.

## Steps

In the sidebar, open **Search Rail** (under **Rails**) and click **Connect source**.

![Connect a data source in Search Rail](https://mintcdn.com/railyai/XX1nxXkekI9EWYGS/images/vector-store/connect-source.png?fit=max\&auto=format\&n=XX1nxXkekI9EWYGS\&q=85\&s=20998c013981b93347a63d1c797a2d46)

Pick where your data lives

Then pick your provider and follow its steps.

> [tab] Qdrant
>
> **1. Open Qdrant**
>
> Click the **Qdrant** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Qdrant URL**, for example `your-project.europe-west3-0.gcp.cloud.qdrant.io:6333`.
> * **API Key**. Required for Qdrant Cloud, optional for local Qdrant.
>
> ![Qdrant connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/qdrant.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=8e0bfb5ec30e672ffa7d7e2538b72dc9)
>
> Qdrant connection form
>
> **3. Add collections**
>
> Click **Fetch Collections** and enable the ones you want, or add a collection by name (lowercase letters, numbers, and underscores).
>
> **4. Set an embedding config**
>
> Pick an embedding config for each collection. It must match the model that created the vectors.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_qdrant.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=2bb749a998d3c0ce305445f16cb6d144)
>
> Collection configuration in the Data Catalog

> [tab] Pinecone
>
> **1. Open Pinecone**
>
> Click the **Pinecone** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **API Key** from your Pinecone console.
>
> ![Pinecone connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/pinecone.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=799f3924d45a7a150c8ac01b9d5aee8c)
>
> Pinecone connection form
>
> **3. Add indexes**
>
> Click **Fetch Indexes** and enable the ones you want. Each enabled index needs its **Index Host URL**, which Raily fills in from Pinecone.
>
> **4. Set an embedding config**
>
> Pick an embedding config for each index. It must match the model that created the vectors. If embeddings are server-generated, set the config to **built-in**.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_pinecone.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=b15f82070b9af3ed8fab0cfdaf92b4a5)
>
> Collection configuration in the Data Catalog

> [tab] Zilliz
>
> **1. Open Zilliz**
>
> Click the **Zilliz** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Cluster endpoint URL**, for example `your-cluster.serverless.gcp-us-west1.cloud.zilliz.com`.
> * **API Key** for your Zilliz Cloud cluster.
>
> ![Zilliz connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/zilliz.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7500db44193414b52352167145e34512)
>
> Zilliz connection form
>
> **3. Add collections**
>
> Click **Fetch Collections** and enable the ones you want, or add a collection by name (start with a letter or underscore; letters, numbers, and underscores).
>
> **4. Set an embedding config**
>
> Pick an embedding config for each collection. It must match the model that created the vectors.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_zilliz.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=ff0a77a48f6031f1dbc6122ad291bdc0)
>
> Collection configuration in the Data Catalog

> [tab] Milvus
>
> **1. Open Milvus**
>
> Click the **Milvus** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Cluster endpoint URL**, for example `milvus.your-company.com`.
> * **API Key**. Optional for self-hosted Milvus.
>
> ![Milvus connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/zilliz.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7500db44193414b52352167145e34512)
>
> Milvus connection form
>
> **3. Add collections**
>
> Click **Fetch Collections** and enable the ones you want, or add a collection by name (start with a letter or underscore; letters, numbers, and underscores).
>
> **4. Set an embedding config**
>
> Pick an embedding config for each collection. It must match the model that created the vectors.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.

> [tab] Weaviate
>
> **1. Open Weaviate**
>
> Click the **Weaviate** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Weaviate URL**, for example `your-cluster.c0.europe-west3.gcp.weaviate.cloud`.
> * **API Key**. Required for Weaviate Cloud, optional if your instance has no authentication.
>
> ![Weaviate connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/weaviate.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=5918aca16a01947e745d9ebc6d94a9e7)
>
> Weaviate connection form
>
> **3. Add collections**
>
> Click **Fetch Collections** and enable the ones you want, or add a collection by name. Weaviate uses PascalCase names, for example `MyCollection`.
>
> **4. Set an embedding config**
>
> Pick an embedding config for each collection. It must match the model that created the vectors. If embeddings are server-generated, set the config to **built-in**.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_weaviate.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=61c26a50abfdaf6066a1438e7c2a54cc)
>
> Collection configuration in the Data Catalog

> [tab] pgvector
>
> **1. Open pgvector**
>
> Click the **pgVector** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Connection String**, for example `postgresql://user@host:5432/dbname`.
> * **Password**. Optional if the password is already in the connection string.
>
> ![pgvector connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/pgvector.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=e9a8c01fe42880cb66d66d5d3b31b956)
>
> pgvector connection form
>
> **3. Confirm the connection**
>
> Raily checks the credentials automatically. The **Credentials** section shows **Verified** when they work, or **Failed** if not.
>
> **4. Add tables**
>
> Click **Fetch Tables** and enable the ones you want, or add a table by name. Schema-qualified names are supported, for example `public.documents`.
>
> **5. Set an embedding config**
>
> Pick an embedding config for each table. It must match the model that created the vectors.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **6. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.

> [tab] Oracle
>
> **1. Open Oracle**
>
> Click the **Oracle** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Connection String** in Easy Connect form, for example `user@host:1521/FREEPDB1`. For Autonomous Database, add `?wallet_location=...`.
> * **Password**. Optional if the password is already in the connection string.
>
> ![Oracle connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/oracle.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=a4ec184d667941a7afde555018cb1244)
>
> Oracle connection form
>
> **3. Confirm the connection**
>
> Raily checks the credentials automatically. The **Credentials** section shows **Verified** when they work, or **Failed** if not.
>
> **4. Add tables**
>
> Click **Fetch Tables** and enable the ones you want, or add a table by name. Schema-qualified names are supported, for example `RAILY.ARTICLES`.
>
> **5. Set an embedding config**
>
> Pick an embedding config for each table. It must match the model that created the vectors. If embeddings are server-generated, set the config to **built-in**.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **6. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **7. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_oracle.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=c4c537ce7e87a490c2c3dad389ee96bb)
>
> Collection configuration in the Data Catalog

> [tab] IBM Ceph
>
> **1. Open IBM Ceph**
>
> Click the **IBM Ceph** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **RGW Endpoint URL** of the Ceph RADOS Gateway, for example `https://ceph.example.com`.
> * **Region**. Required by the S3 client; Ceph ignores it. `us-east-1` is fine.
> * **Object Bucket** where uploaded files are stored. Vectors live in `<bucket>-vectors`.
> * **Index Name**. Use `default` unless you have multiple indexes per bucket.
> * **Access Key ID** and **Secret Access Key** for the Ceph cluster.
>
> ![IBM Ceph connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/ceph.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=6299a6dacbfed737553a56bede44fcf9)
>
> IBM Ceph connection form
>
> **3. Add vector buckets**
>
> Click **Fetch Vector Buckets** and enable the ones you want.
>
> **4. Set an embedding config**
>
> Pick an embedding config for each vector bucket. It must match the model that created the vectors.
>
> Don't have one yet? Open the embedding config picker and click **Create New**.
>
> ![Embedding config picker with Create New](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create1.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=61879b413bbc70398a16893cb781b2fc)
>
> Create a new embedding config
>
> Enter a **Name**, pick a **Provider**, and set the **Model Name** and **Dimension** to match the model that created your vectors. Add an **API Key** if the provider needs one, then save.
>
> ![Embedding config form fields](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/create2.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=7cd576be83273091a9c9f0f47df8a244)
>
> Embedding config form
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_ceph.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=8d04973b5f3444cfdf3f23a3722b6d69)
>
> Collection configuration in the Data Catalog

> [tab] Salesforce
>
> **1. Open Salesforce**
>
> Click the **Salesforce** card.
>
> **2. Enter the connection details**
>
> * **Name** for the integration.
> * **Instance URL**, for example `https://myorg.my.salesforce.com`.
> * **Client ID**, your Connected App consumer key.
> * **Client Secret**, your Connected App consumer secret.
>
> ![Salesforce connection form](https://mintcdn.com/railyai/LT7rWl_muWq2wYsL/images/vector-store/salesforce.png?fit=max\&auto=format\&n=LT7rWl_muWq2wYsL\&q=85\&s=ca70563d105b5627eb00b0e371822a55)
>
> Salesforce connection form
>
> **3. Confirm the connection**
>
> Raily checks the credentials automatically. The **Credentials** section shows **Verified** when they work, or **Failed** if not.
>
> **4. Add collections**
>
> Click **Fetch Collections** and enable the ones you want. Salesforce search runs over Cases via SOSL, so no embedding config is needed.
>
> **5. Save**
>
> Click **Create Integration**. Raily then configures agentic search for each collection automatically. This can take a few minutes, after which the integration appears under **My Integrations** with an **Active** status.
>
> > [!NOTE]
> >
> > Once it finishes, open **Inventory → Data Catalog** to see the generated configuration for each collection.
>
> **6. See it in the Data Catalog**
>
> After generation finishes, Raily opens the collection in **Data Catalog**.
>
> ![Collection configuration in the Data Catalog](https://mintcdn.com/railyai/jfCeh5-7enF3hO2p/images/vector-store/data_catalog_saleforce.png?fit=max\&auto=format\&n=jfCeh5-7enF3hO2p\&q=85\&s=9dab4b5a53769498405ec10bafb0dcf3)
>
> Collection configuration in the Data Catalog

> [tab] Raily Vector Store
>
> Don't have a vector database yet? [Talk to us](https://www.raily.ai/contact) and we can provide one for you.

## Best practices

> [card] Match the embedding config
>
> The embedding config must match the model that created the vectors.

> [card] Design useful metadata
>
> Index the fields people filter by so the agent can use them.

> [card] Keep embeddings current
>
> Re-embed when content changes so search stays accurate.

> [card] Use access control
>
> Apply policies so each endpoint exposes only what it should.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.