PgVector Vector Database

Use PgVector as a vector database for your Knowledge Base.

Setup

1docker run -d \
2  -e POSTGRES_DB=ai \
3  -e POSTGRES_USER=ai \
4  -e POSTGRES_PASSWORD=ai \
5  -e PGDATA=/var/lib/postgresql/data/pgdata \
6  -v pgvolume:/var/lib/postgresql/data \
7  -p 5532:5432 \
8  --name pgvector \
9  agnohq/pgvector:16

Example

1from kern.agent import Agent
2from kern.models.openai import OpenAIResponses
3from kern.knowledge.knowledge import Knowledge
4from kern.vectordb.pgvector import PgVector, SearchType
5
6db_url = "postgresql+psycopg://ai:ai@localhost:5532/ai"
7knowledge_base = Knowledge(
8    vector_db=PgVector(table_name="recipes", db_url=db_url, search_type=SearchType.hybrid),
9)
10
11if __name__ == "__main__":
12    knowledge_base.insert(
13        url="https://kern-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
14    )
15
16    agent = Agent(
17        model=OpenAIResponses(id="gpt-5.2"),
18        knowledge=knowledge_base,
19        # Add a tool to read chat history.
20        read_chat_history=True,
21        markdown=True,
22        # debug_mode=True,
23    )
24    agent.print_response("How do I make chicken and galangal in coconut milk soup", stream=True)
25    agent.print_response("What was my last question?", stream=True)

Async Support ⚡

PgVector also supports asynchronous operations, enabling concurrency and leading to better performance.

1import asyncio
2
3from kern.agent import Agent
4from kern.knowledge.knowledge import Knowledge
5from kern.vectordb.pgvector import PgVector
6
7db_url = "postgresql+psycopg://ai:ai@localhost:5532/ai"
8
9vector_db = PgVector(table_name="recipes", db_url=db_url)
10
11knowledge_base = Knowledge(
12    vector_db=vector_db,
13)
14
15agent = Agent(knowledge=knowledge_base)
16
17if __name__ == "__main__":
18    # Load knowledge base asynchronously
19    asyncio.run(knowledge_base.ainsert(
20            url="https://kern-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
21        )
22    )
23
24    # Create and use the agent asynchronously
25    asyncio.run(agent.aprint_response("How to make Tom Kha Gai", markdown=True))

Tip

Use aload() and aprint_response() methods with asyncio.run() for non-blocking operations in high-throughput applications.

PgVector Params

Parameter	Type	Default	Description
`table_name`	`str`	-	The name of the table to use.
`schema`	`str`	-	The schema to use.
`db_url`	`str`	-	The database URL to connect to.
`db_engine`	`Engine`	-	The database engine to use.
`embedder`	`Embedder`	-	The embedder to use.
`search_type`	`SearchType`	vector	The search type to use.
`vector_index`	`Union[Ivfflat, HNSW]`	-	The vector index to use.
`distance`	`Distance`	cosine	The distance to use.
`prefix_match`	`bool`	-	Whether to use prefix matching.
`vector_score_weight`	`float`	0.5	Weight for vector similarity in hybrid search. Must be between 0 and 1.
`content_language`	`str`	-	The content language to use.
`schema_version`	`int`	-	The schema version to use.
`auto_upgrade_schema`	`bool`	-	Whether to auto upgrade the schema.