Your ideas.
All the building blocks.

Version your prompts, reuse shared instructions, and compare evaluations before choosing what runs in your application.

• Versioned prompts and reusable parts
• Model configurations and structured outputs
• Execution history and evaluations

Connect models, tools, and decisions in visual workflows. Keep the context of a run as it waits, resumes, and moves forward.

• Visual workflow authoring
• Versioned execution bundles
• Durable waits, retries, and run history

Bring together instructions, conversation context, and tools. Keep people involved with explicit approvals when an action needs a decision.

• Persistent threads and conversation history
• Tool calls and human approvals
• Streaming responses and session context

Organize documents and retrieve relevant context so your AI application can work with the information that matters to your team.

• Document storage and collections
• Vector search and retrieval
• Context for prompts and agents

Connect application tools and model providers. Bring LangStack capabilities into your own products and agent environments through APIs, SDKs, and MCP.

• Reusable tools and provider connections
• Java and TypeScript integrations
• Workspace-scoped MCP access

Follow executions from input to output, inspect failures, and review available latency and token usage alongside your prompt history.

• Workflow traces and execution details
• Prompt performance and usage metadata
• Shared context within your workspace