# Pixeltable

Pixeltable is declarative data infrastructure for multimodal AI. It unifies storage, orchestration, and retrieval under one table interface — so you can build production-ready video, image, audio, and document workflows in minutes, not months. Open source, pip-installable, no separate vector databases or ETL pipelines required.

## Getting Started

- [Homepage](https://pixeltable.com): Product overview and live demos
- [Quick Start](https://docs.pixeltable.com/overview/quick-start): Install with `pip install pixeltable` and run your first pipeline
- [10-Minute Tour](https://docs.pixeltable.com/overview/ten-minute-tour): Hands-on walkthrough of core features
- [Configuration](https://docs.pixeltable.com/platform/configuration): Environment setup, API keys, and storage options
- [Starter Kit](https://github.com/pixeltable/pixeltable-starter-kit): Full-stack multimodal AI app template — fork and ship

## Core Concepts

- [Tables & Data Operations](https://docs.pixeltable.com/tutorials/tables-and-data-operations): Create tables with native multimodal types (`pxt.Video`, `pxt.Image`, `pxt.Audio`, `pxt.Document`)
- [Computed Columns](https://docs.pixeltable.com/tutorials/computed-columns): Declarative AI pipelines — define once, auto-execute on insert
- [Queries & Expressions](https://docs.pixeltable.com/tutorials/queries-and-expressions): SQL-like `.where()`, `.select()`, `.order_by()`, `.similarity()` in Python
- [Views & Iterators](https://docs.pixeltable.com/platform/views): Virtual tables for frame extraction, chunking, and derived data
- [Iterators](https://docs.pixeltable.com/platform/iterators): Split documents, extract frames, chunk audio
- [Embedding Indexes](https://docs.pixeltable.com/platform/embedding-indexes): One-liner vector search with any embedding model
- [Custom Functions (UDFs)](https://docs.pixeltable.com/platform/udfs-in-pixeltable): Wrap any Python function with `@pxt.udf`
- [Type System](https://docs.pixeltable.com/platform/type-system): Native multimodal types for structured data, media, and ML
- [Version Control](https://docs.pixeltable.com/platform/version-control): Every table is transactional and fully versioned — `t.revert()`, `t.history()`
- [Data Sharing](https://docs.pixeltable.com/platform/data-sharing): Publish and replicate datasets via Pixeltable Cloud

## AI Integrations

- [OpenAI](https://docs.pixeltable.com/howto/providers/working-with-openai): GPT-4o, Whisper, DALL-E, embeddings
- [Anthropic](https://docs.pixeltable.com/howto/providers/working-with-anthropic): Claude models for chat and analysis
- [Google Gemini](https://docs.pixeltable.com/howto/providers/working-with-gemini): Gemini multimodal models, Imagen, Veo
- [Hugging Face](https://docs.pixeltable.com/howto/providers/working-with-hugging-face): Thousands of open-source models (CLIP, DETR, Whisper, etc.)
- [Twelve Labs](https://docs.pixeltable.com/howto/providers/working-with-twelvelabs): Cross-modal video embeddings
- [Together AI](https://docs.pixeltable.com/howto/providers/working-with-together): Open-source model inference
- [Fireworks](https://docs.pixeltable.com/howto/providers/working-with-fireworks): Fast inference for open models
- [Ollama](https://docs.pixeltable.com/howto/providers/working-with-ollama): Local model inference
- [All Integrations](https://docs.pixeltable.com/integrations/frameworks): Complete list of 30+ supported AI providers and frameworks

## Use Cases

- [Backend for AI Apps](https://docs.pixeltable.com/use-cases/ai-applications): RAG pipelines, semantic search, and multimodal APIs
- [Data Wrangling for ML](https://docs.pixeltable.com/use-cases/ml-data-wrangling): Ingest, annotate, and export multimodal training data
- [Agents & MCP](https://docs.pixeltable.com/use-cases/agents-mcp): Tool calling, persistent memory, MCP server integration

## Deployment

- [Deployment Overview](https://docs.pixeltable.com/howto/deployment/overview): Two approaches — full backend or ephemeral orchestration
- [Infrastructure Setup](https://docs.pixeltable.com/howto/deployment/infrastructure): Code organization, storage architecture, and project structure
- [Production Operations](https://docs.pixeltable.com/howto/deployment/operations): Concurrency, error handling, schema evolution, and FastAPI patterns
- [Security & Backup](https://docs.pixeltable.com/howto/deployment/security): Backup strategies, recovery, and security best practices

## Cookbooks

- [RAG Pipeline](https://docs.pixeltable.com/howto/cookbooks/agents/pattern-rag-pipeline): Document chunking, embedding, retrieval, and LLM generation
- [Tool Calling](https://docs.pixeltable.com/howto/cookbooks/agents/llm-tool-calling): Register UDFs as LLM tools, MCP server integration
- [Agent Memory](https://docs.pixeltable.com/howto/cookbooks/agents/pattern-agent-memory): Persistent conversation context with semantic recall
- [Agentic Patterns](https://docs.pixeltable.com/howto/cookbooks/agents/agentic-patterns): All 8 agentic patterns as Pixeltable tables
- [Video Frame Extraction](https://docs.pixeltable.com/howto/cookbooks/video/video-extract-frames): Extract frames at fixed rate, keyframes, or exact count
- [Audio Transcription](https://docs.pixeltable.com/howto/cookbooks/audio/audio-transcribe): Transcribe audio with Whisper
- [Document Chunking](https://docs.pixeltable.com/howto/cookbooks/text/doc-chunk-for-rag): Split documents for RAG
- [Semantic Text Search](https://docs.pixeltable.com/howto/cookbooks/search/search-semantic-text): Search by meaning, not keywords
- [Similar Image Search](https://docs.pixeltable.com/howto/cookbooks/search/search-similar-images): Visual similarity with CLIP
- [Object Detection](https://docs.pixeltable.com/howto/cookbooks/images/img-detect-objects): Detect objects with YOLOX
- [Data Import (CSV, JSON, Parquet, S3)](https://docs.pixeltable.com/howto/cookbooks/data/data-import-csv): Load data from any source
- [Data Export (SQL, PyTorch, Parquet)](https://docs.pixeltable.com/howto/cookbooks/data/data-export-sql): Export to databases and ML formats

## SDK & API Reference

- [SDK Reference](https://docs.pixeltable.com/sdk/latest/pixeltable): Core functions — `create_table()`, `tools()`, `udf()`, `query()`
- [Table API](https://docs.pixeltable.com/sdk/latest/table): `insert()`, `select()`, `where()`, `add_computed_column()`, `add_embedding_index()`
- [Query API](https://docs.pixeltable.com/sdk/latest/query): `collect()`, `order_by()`, `limit()`, `join()`, `group_by()`
- [Built-in Functions](https://docs.pixeltable.com/sdk/latest/functions): Aggregates (`sum`, `count`, `mean`, `min`, `max`) and utilities
- [Image Functions](https://docs.pixeltable.com/sdk/latest/image): `resize()`, `rotate()`, `crop()`, `convert()`, and more
- [Video Functions](https://docs.pixeltable.com/sdk/latest/video): `frame_iterator()`, `video_splitter()`, `extract_audio()`, `make_video()`
- [I/O Functions](https://docs.pixeltable.com/sdk/latest/io): `import_csv()`, `export_parquet()`, `export_lancedb()`

## Developer Tools

- [Building with LLMs](https://docs.pixeltable.com/overview/building-pixeltable-with-llms): Guide for AI coding assistants working with Pixeltable
- [MCP Server](https://github.com/pixeltable/mcp-server-pixeltable-developer): Model Context Protocol server for Claude, Cursor, and AI IDEs
- [AGENTS.md](https://github.com/pixeltable/pixeltable/blob/main/AGENTS.md): Architecture guide for AI agents working with the codebase
- [llms-full.txt](https://docs.pixeltable.com/llms-full.txt): Complete documentation optimized for LLM consumption

## Open Source

- [GitHub](https://github.com/pixeltable/pixeltable): Main repository — star, fork, contribute
- [Pixelagent](https://github.com/pixeltable/pixelagent): Stateful AI agent framework built on Pixeltable
- [Pixelbot](https://github.com/pixeltable/pixelbot): Infinite-memory chat application
- [Changelog](https://docs.pixeltable.com/changelog/changelog): Release history and updates

## Migration Guides

- [From DIY Data Pipelines](https://docs.pixeltable.com/migrate/from-diy-data-pipeline): Replace custom scripts, DVC, Airflow
- [From RDBMS & Vector DBs](https://docs.pixeltable.com/migrate/from-rdbms-vectordbs): Replace Postgres + Pinecone + LangChain
- [From Agent Frameworks](https://docs.pixeltable.com/migrate/from-agent-frameworks): Replace LangGraph, CrewAI

## Pixeltable Cloud

- [Cloud Platform](https://pixeltable.com/dashboard): Managed Pixeltable with collaboration and sharing
- [Public Datasets](https://pixeltable.com/data-products): Browse and replicate shared multimodal datasets
- [Pricing](https://pixeltable.com/pricing): Free tier and paid plans
- [Cloud Offering](https://docs.pixeltable.com/use-cases/services): Data sharing, endpoints, and collaboration

## Community

- [Discord](https://discord.gg/QPyqFYx2UN): Community chat and support
- [YouTube](https://www.youtube.com/@PixeltableHQ): Demos, talks, and tutorials
- [Blog](https://pixeltable.com/blog): Technical articles and product updates
- [X / Twitter](https://twitter.com/pixeltablehq): Product updates

## Contact

- Website: [https://pixeltable.com](https://pixeltable.com)
- Documentation: [https://docs.pixeltable.com](https://docs.pixeltable.com)
- GitHub: [https://github.com/pixeltable/pixeltable](https://github.com/pixeltable/pixeltable)
- Company: Pixeltable, Inc.

## Last updated

March 2026