Skip to main content

​
Create Vector Database

Create a new vector database configuration to store and query high-dimensional vectors for semantic search.

​
Request Body

​
provider
string
required
Vector database provider:
  • PINECONE - Pinecone vector database service
  • ASTRA_DB - DataStax Astra DB vector search
  • WEAVIATE - Weaviate vector database
  • QDRANT - Qdrant vector search engine
  • SUPABASE - Supabase with pgvector extension
​
options
object
required
Provider-specific configuration options (see provider sections below for details)

​
Response

Returns the created vector database configuration.
​
id
string
Vector database configuration unique identifier
​
provider
string
Vector database provider
​
options
string
Encrypted configuration options
​
project_id
string
ID of the project this vector database belongs to
​
created_at
string
When the configuration was created (ISO 8601 format)
​
updated_at
string
When the configuration was last updated (ISO 8601 format)
curl --location --request POST 'https://api.plaisolutions.com/vector-db' \
--header 'Authorization: Bearer YOUR_TOKEN' \
--header 'Content-Type: application/json' \
--data-raw '{
  "provider": "PINECONE",
  "options": {
    "api_key": "your-pinecone-api-key",
    "environment": "us-west1-gcp-free",
    "index_name": "plai-vectors"
  }
}'

​
List Vector Databases

List all vector database configurations in the current project.

​
Response

​
vector_dbs
array
Array of vector database configuration objects
curl --location --request GET 'https://api.plaisolutions.com/vector-dbs' \
--header 'Authorization: Bearer YOUR_TOKEN'

​
Get Vector Database

Get detailed information about a specific vector database configuration.

​
Path Parameters

​
vector_db_id
string
required
The unique identifier of the vector database configuration

​
Response

Returns the vector database configuration object with all details. 
curl --location --request GET 'https://api.plaisolutions.com/vector-dbs/vectordb_123' \
--header 'Authorization: Bearer YOUR_TOKEN'

​
Update Vector Database

Update vector database configuration settings.

​
Path Parameters

​
vector_db_id
string
required
The unique identifier of the vector database configuration

​
Request Body

​
provider
string
Updated vector database provider
​
options
object
Updated provider-specific configuration options

​
Response

Returns the updated vector database configuration object. 
curl --location --request PATCH 'https://api.plaisolutions.com/vector-dbs/vectordb_123' \
--header 'Authorization: Bearer YOUR_TOKEN' \
--header 'Content-Type: application/json' \
--data-raw '{
  "options": {
    "api_key": "new-pinecone-api-key",
    "environment": "us-east1-gcp",
    "index_name": "plai-vectors-prod"
  }
}'

​
Provider Configuration

Each vector database provider requires specific configuration options:

​
Pinecone

{
  "api_key": "your-pinecone-api-key",
  "environment": "us-west1-gcp-free",
  "index_name": "your-index-name",
  "namespace": "default"
}
  1. Create a Pinecone account at pinecone.io
  2. Create a new project and get your API key
  3. Create an index with the desired dimensions (typically 1536 for OpenAI embeddings)
  4. Choose the appropriate environment/region
  5. Optionally specify a namespace for data isolation
  • Fully managed vector database
  • High-performance approximate nearest neighbor search
  • Real-time updates and deletes
  • Metadata filtering
  • Horizontal scaling

​
DataStax Astra DB

{
  "token": "your-astra-db-token",
  "api_endpoint": "https://your-database-id-region.apps.astra.datastax.com",
  "keyspace": "your-keyspace",
  "collection": "your-collection"
}
  1. Create an Astra DB account at astra.datastax.com
  2. Create a new vector database
  3. Generate an application token
  4. Create a keyspace and collection
  5. Configure the collection for vector search
  • Serverless vector database built on Cassandra
  • Multi-cloud availability
  • ACID transactions
  • JSON document model with vector search
  • Global distribution capabilities

​
Weaviate

{
  "url": "https://your-cluster.weaviate.network",
  "api_key": "your-weaviate-api-key",
  "class_name": "Document",
  "additional_headers": {
    "X-OpenAI-Api-Key": "your-openai-key"
  }
}
  1. Create a Weaviate cluster (cloud or self-hosted)
  2. Configure authentication if required
  3. Create schema classes for your data
  4. Set up vectorization modules (OpenAI, Cohere, etc.)
  5. Configure any additional modules needed
  • Open-source vector database
  • GraphQL and REST APIs
  • Built-in vectorization modules
  • Semantic search and hybrid search
  • Multi-modal support (text, images)

​
Qdrant

{
  "url": "https://your-cluster.qdrant.io",
  "api_key": "your-qdrant-api-key",
  "collection_name": "plai_vectors",
  "vector_size": 1536,
  "distance": "Cosine"
}
  1. Set up Qdrant (cloud or self-hosted)
  2. Create a collection with appropriate vector size
  3. Configure distance metric (Cosine, Euclidean, Dot product)
  4. Set up authentication if required
  5. Optionally configure payload indexing
  • High-performance vector similarity search
  • Rich payload support with filtering
  • Distributed deployment
  • Real-time updates
  • Advanced indexing algorithms (HNSW)

​
Supabase

{
  "url": "https://your-project.supabase.co",
  "key": "your-anon-key",
  "table_name": "documents",
  "vector_column": "embedding",
  "content_column": "content",
  "metadata_column": "metadata"
}
  1. Create a Supabase project
  2. Enable the pgvector extension in your database
  3. Create tables with vector columns
  4. Set up Row Level Security (RLS) policies
  5. Configure API keys and authentication
  • PostgreSQL with pgvector extension
  • Full SQL capabilities with vector operations
  • Row-level security
  • Real-time subscriptions
  • Built-in authentication and storage

​
Best Practices

1

Choose the Right Provider

Select a vector database provider based on your performance, scalability, and budget requirements
2

Dimension Consistency

Ensure all vectors have the same dimensions as configured in your vector database
3

Index Configuration

Configure appropriate index parameters for your expected query patterns and data size
4

Security

Store API keys and credentials securely, never expose them in client-side code
5

Monitoring

Monitor query performance and storage usage to optimize costs and performance

​
Vector Database Selection Guide

High Performance

Pinecone or Qdrant for maximum query speed and throughput

Cost Effective

Supabase with pgvector for budget-conscious projects with SQL needs

Enterprise

Astra DB for enterprise features, global distribution, and ACID compliance

Flexibility

Weaviate for multi-modal search and complex schema requirements

​
Performance Optimization

  • Use appropriate vector dimensions (768, 1536, 3072 are common)
  • Implement efficient batch operations for bulk updates
  • Consider using namespaces or collections to organize data
  • Monitor and tune index parameters based on your query patterns

​
Security Considerations

Vector databases often contain sensitive information:
  • Use encrypted connections (HTTPS/TLS)
  • Implement proper access controls and authentication
  • Regularly rotate API keys and credentials
  • Monitor access logs for unusual activity
  • Consider data residency requirements for compliance

​
Common Use Cases

Configure your vector database to power semantic search across documents, enabling users to find content based on meaning rather than exact keyword matches.

​
Recommendation Systems

Use vector similarity to build recommendation engines that suggest similar products, content, or services based on user behavior and preferences.

​
Retrieval-Augmented Generation (RAG)

Combine your vector database with language models to provide contextually relevant information for AI-powered applications.

​
Duplicate Detection

Identify duplicate or near-duplicate content by finding vectors with high similarity scores within your dataset.

​
Clustering and Classification

Group related items together or classify new content by finding the closest matching vectors in your database.