Document ingestion for AI builders

Your knowledge base,
ready for AI search.

Connect your documents, prepare searchable content, and load it into your vector database. Test retrieval and keep supported sources updated automatically.

Start building freeSee how it works
  • 250 one-time startup credits
  • No credit card required
  • 12 source types

  1. 1. Choose documents

    Connect sources or upload files

  2. 2. Prepare and sync

    Extract, chunk, and embed

  3. 3. Store vectors

    Use your configured database

  4. 4. Build with your data

    Test retrieval and connect your app

Interactive product preview · sample data

See It in Action

Explore the product workflow. Switch views, browse sample documents, and try a sample search.

Sample workspace

Command Center

Sample operations overview

1,247

Total Documents

45.2K

Vector Chunks

6

Active Sources

23

Syncs Today

Recent Activity

Engineering Wiki / API Docs

Confluence

Queued

Product Roadmap Q4

Google Drive

Synced

Support Knowledge Base

Confluence

Synced

Marketing Assets

Google Drive

Synced

Illustrative product preview. Figures and documents are sample data; no live connections or syncs are made.

How It Works

From Raw Data to Vector Search

Connect, prepare, and search your documents — with your choice of embedding provider and vector store.

Connect Your Sources to Your Vector Store

Connect supported sources with OAuth or provider credentials. We parse, chunk, and embed selected documents into your configured vector database.

Pinecone
pgvector
6 stores supported

Smart Chunking

Structure-aware

Intro
chunk_1
Section A
chunk_2
Section B
chunk_3
Section C
chunk_4

Structure-aware splitting keeps supported headings and source context with each chunk.

Auto-Sync Engine

Staleness detection

00:00
API Docs
02:14
Release Notes
now
Wiki Changes

Schedule document checks. Unchanged documents skip re-embedding; changed documents are processed again.

4 Embedding Providers

Plug in your preferred model

OpenAI, Cohere, Google Gemini, or a reachable Ollama service. Model changes may require reprocessing and a compatible vector-store configuration.

OpenAI
Cohere
Google Gemini
Ollama

Built-in Playground

Test retrieval against your documents

How do I configure webhooks?
Webhook Guide94%
API Reference89%
Illustrative results · not a performance benchmark
SSRF-protected web crawler
Automatic retry on failure
Metadata-enriched chunks
Project-level isolation

Choose the pieces that fit your project.

Connect supported sources to your configured vector store and embedding provider. Each project can have its own configuration.

12 source types
6 vector stores
4 embedding providers
Compare setup and capabilities

Pay as you go

Start free. Buy credits when you need them.

250 one-time startup credits cover 12 standard document syncs. Purchased credits never expire.

Try the full workflow
  • Supported source connectors and vector stores
  • Project-specific configurations
  • Search and chat playground
  • No credit card required

Configure your embedding provider and vector store. Their charges are separate; optional LLM/OCR processing may also require your own provider credentials.

Start building free
Starter

$5.00 one-time

500 credits

Standard

$25.00 one-time

3,000 credits

Value

$50.00 one-time

7,500 credits

Bulk

$100.00 one-time

18,000 credits

Standard sync: 20 credits. Re-syncs and stored uploads/images can use additional credits. See all costs and estimate your workload.

Ready to try your first project

Stop writing ETL scripts.
Start building AI.

Prepare documents from your supported sources, test retrieval, and connect the results to your AI application.

Start Building FreeContact Sales
DocumentationNo credit card required to start