ByteChef

ByteChef is an open-source, self-hosted web app and server that combines AI agent orchestration with workflow automation. It provides a unified platform where AI agents and automated sequences operate within a single orchestration layer, allowing users to build complex operations without extensive coding. The platform aims to replace the need for separate tools by bringing agent loops and linear automation into one centralized management system.

The software is deployed as a server application, typically via Docker or Kubernetes. It features a visual drag-and-drop editor for designing workflows and agents, which can be triggered by webhooks, schedules, polling, or manual inputs. It is designed for use in regulated environments and can be embedded into other products to provide AI capabilities to end users. Users can build agents by selecting a model, attaching tools from a library of connectors, and adding knowledge bases or guardrails.

Key features

• Drag-and-drop AI Agent component with support for 14 LLM providers
• Integration with 15+ vector stores for knowledge bases and RAG patterns
• Support for 12 guardrails including PII detection and jailbreak prevention
• MCP support to consume MCP servers as tools or expose workflows as MCP tools
• Polyglot code execution supporting Java, JavaScript, Python, and Ruby
• Durable execution backed by Postgres with queue-mode for horizontal scaling
• Over 180 connectors for CRM, e-commerce, cloud storage, and databases
• Human-in-the-loop capabilities to pause workflows for approval or response

ByteChef utilizes the Atlas runtime for execution and supports multiple memory backends including Redis, MongoDB, Cassandra, and Neo4j. The architecture allows agents to function as steps within a workflow, while entire workflows can be treated as tools for an agent to call. This bidirectional relationship enables the creation of coordinator agents that manage specialist sub-agents. The platform is intended for enterprises requiring centralized management and audit logs that track agent decisions, tool calls, and human approvals in a single trail. It also supports various queue modes using RabbitMQ, Kafka, or SQS for scaling.

ByteChef is an open-core platform that bridges the gap between traditional iPaaS automation and modern AI agent frameworks.