Vector Search

Find content by semantic meaning using vector similarity.

Vector search finds content by meaning rather than exact word matches. When you search for “How do I reset my password?”, it finds documents about “changing credentials” even if those exact words don’t appear.

from agno.knowledge.knowledge import Knowledge
from agno.vectordb.pgvector import PgVector, SearchType

knowledge = Knowledge(
    vector_db=PgVector(
        table_name="docs",
        db_url=db_url,
        search_type=SearchType.vector,
    ),
)

How It Works

1. Query embedding: Your search query is converted to a vector (list of numbers capturing meaning)
2. Similarity matching: The system finds stored vectors closest to the query vector
3. Ranking: Results are ordered by cosine similarity (how close the meanings are)

The embedding model determines how well semantic relationships are captured. General-purpose models like OpenAI’s text-embedding-3-small work well for most content.

When to Use Vector Search

Scenario	Why Vector Search Works
Conceptual questions	Matches meaning, not just words
Users phrase things differently	Finds relevant content regardless of terminology
Natural language queries	Understands intent behind questions
Content with varied vocabulary	Connects synonyms and related concepts

Use hybrid search if you also need exact term matching (product names, error codes). Use keyword search if you need precise text matching only.

Configuration

Basic Setup

from agno.knowledge.embedder.openai import OpenAIEmbedder

vector_db = PgVector(
    table_name="docs",
    db_url=db_url,
    search_type=SearchType.vector,
    embedder=OpenAIEmbedder(id="text-embedding-3-small"),
)

With Reranking

Add a reranker to improve result ordering:

from agno.knowledge.reranker.cohere import CohereReranker

vector_db = PgVector(
    table_name="docs",
    db_url=db_url,
    search_type=SearchType.vector,
    reranker=CohereReranker(),
)

Example

vector_search.py

from agno.knowledge.knowledge import Knowledge
from agno.vectordb.pgvector import PgVector, SearchType

db_url = "postgresql+psycopg://ai:ai@localhost:5532/ai"

knowledge = Knowledge(
    vector_db=PgVector(
        table_name="recipes",
        db_url=db_url,
        search_type=SearchType.vector,
        embedder=OpenAIEmbedder(id="text-embedding-3-small"),
    ),
)

# Load content
knowledge.insert(
    url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf",
)

# Search by semantic meaning
results = knowledge.search("chicken coconut soup", max_results=5)
for doc in results:
    print(doc.content[:200])

Next Steps

Hybrid Search

Combine vector search with keyword matching

Embedders

Choose the right embedding model