Local-first retrieval.Built in.

Embedded retrieval for applications. Local workspace search for developers and AI agents.

PYTHON SDK

schema = zvec.CollectionSchema(
    name="docs",
    vectors=zvec.VectorSchema(
        "embedding", zvec.DataType.VECTOR_FP32, 768
    ),
)
collection = zvec.create_and_open(
    path="./zvec_data", schema=schema
)

Products across retrieval workflows

Zvec embeds retrieval in applications; Zvec-Grep brings local workspace search to people and AI agents.

EMBEDDED ENGINE

Zvec

Dense, sparse, full-text, and hybrid retrieval embedded directly in your application.

Explore Zvec

SDK & APIs

Python Node.js Go Rust Dart C / C++

Search & indexes

Retrieval

INDEX TYPES

Quantization

Runtime & platform

INFRASTRUCTURE

RUNTIMES

LinuxmacOSWindowsAndroidiOS

CLI & MCP

Zvec-Grep

Local search beyond exact keywords—for people and AI agents.

Explore Zvec-Grep

CODEMDTXTDATA

exactBM25vectorhybrid

Ranked evidencepaths · symbols · lines

PeopleAI agents

Beyond keywordsSource-linkedLocal CLI + MCP

SWE-QA AGENT BENCHMARK

+1.5 ppquality−47%tokens−59%tool calls−38%agent time

Open-source integrations

Beyond the product family, open-source projects embed Zvec across knowledge tools, agent memory, AI frameworks, and data pipelines.

Knowledge management

Logseq

Desktop hybrid search over local graphs and embeddings.

Zvec roleHybrid retrieval

Personal context

MyContext

Private context retrieval and knowledge-graph memory.

Zvec roleVector storage

Agent memory

ReMe

Persistent local memory that agents can search and recall.

Zvec roleMemory backend

AI framework

txtai

Semantic search inside an all-in-one AI framework.

Zvec roleDense ANN

Data pipeline

CocoIndex

Incremental data pipelines that stay connected to retrieval.

Zvec roleIndex connector