All integrations
đź§ 
AI Agent Framework

LangGraph + Norbix

Stateful AI agents with a real backend.

How it works

1

Give your graph Norbix as a tool node

Wrap the SDK calls your nodes need as tools — Python (pip install norbix) or TypeScript (@norbix.ai/ts) — or call the REST API, signed with a service-user (nbsu_) key. An MCP client pointed at https://hub.norbix.ai/v3/account/mcp with the same key works too. There is no separate Norbix package to install.

2

Define graph nodes that interact with modules

Build stateful graphs where nodes query data, create records, manage users, and process payments. LangGraph maintains state across decisions.

3

Run stateful agents against production

Deploy agents that reason across multiple steps, maintain conversation context, and act on your Norbix infrastructure.

What LangGraph can do with Norbix

Data queries in graph nodes

Query your database, fetch user records, check permissions within agent reasoning steps.

User management

Create users, manage roles and permissions, handle authentication and invitations in workflows.

Payment orchestration

Process charges, manage subscriptions, handle refunds based on agent decisions.

Notification generation

Trigger emails, SMS, and in-app notifications based on agent reasoning and actions.

Multi-step reasoning

Use LangGraph state to remember decisions across steps and interact with backend accordingly.

Conditional backend actions

Agents decide which Norbix actions to take based on data, conditions, and reasoning.

Try LangGraph + Norbix

Create a free account, make a service-user key, and deploy your first stateful agent.