Native MCP + AI Assistant

AI tools that work on real data. Under your rules.

The AI Assistant works inside your Studio. The MCP server connects any external tool to your backend. The same role-based permissions govern both.

Native MCP + AI Assistant

AI tools that work on real data. Under your rules.

The AI Assistant works inside your Studio. The MCP server connects any external tool to your backend. The same role-based permissions govern both.

Native MCP + AI Assistant

AI tools that work on real data. Under your rules.

The AI Assistant works inside your Studio. The MCP server connects any external tool to your backend. The same role-based permissions govern both.

Trusted by teams at

Inside Directus

Conversational AI, native to your data.

The AI Assistant is built into the Studio. Writes, edits, and updates your data within your role's permissions.

Feature

1 / 4

Edit Records, Not Just Content

Ask the assistant to update customer accounts, adjust inventory, modify access policies, or edit content across any collection. Every change goes through your schema and permissions.

Translate Across Languages

Explore and Manage Your Schema

Trigger Automations and Workflows

1 / 4

Edit Records, Not Just Content

Ask the assistant to update customer accounts, adjust inventory, modify access policies, or edit content across any collection. Every change goes through your schema and permissions.

1 / 4

Edit Records, Not Just Content

Ask the assistant to update customer accounts, adjust inventory, modify access policies, or edit content across any collection. Every change goes through your schema and permissions.

Outside Directus

A secure access layer for AI and agents.

Connect Claude, Cursor, ChatGPT, or any MCP-compatible tool directly to your backend. Same permissions without middleware or data exports.

Outside Directus

A secure access layer for AI and agents.

The native MCP server gives your AI tools governed, direct access to your Directus backend. No middleware or data exports needed.

Outside Directus

A secure access layer for AI and agents.

The native MCP server gives your AI tools governed, direct access to your Directus backend. No middleware or data exports needed.

Governance

One permissions system for both humans and AI.

AI agents get a Directus role, just like every other user. Field-level permissions. Collection-level access. The same policy layer that governs your team governs your AI.

Every action is logged

The audit trail doesn't distinguish between a human editor and an AI agent. Who changed what, when, through which tool.

Your data stays in your database

The AI Assistant and MCP server operate on your backend. Content isn't copied to a third-party service for processing.

Control Read/Write Access

Decide what your AI agents can read, write, or never touch. Set the rules per field, per collection.

Governance

One permissions system for both humans and AI.

AI agents get a Directus role, just like every other user. Field-level permissions. Collection-level access. The same policy layer that governs your team governs your AI.

Every action is logged

The audit trail doesn't distinguish between a human editor and an AI agent. Who changed what, when, through which tool.

Your data stays in your database

The AI Assistant and MCP server operate on your backend. Content isn't copied to a third-party service for processing.

Control Read/Write Access

Decide what your AI agents can read, write, or never touch. Set the rules per field, per collection.

Governance

One permissions system for both humans and AI.

AI agents get a Directus role, just like every other user. Field-level permissions. Collection-level access. The same policy layer that governs your team governs your AI.

Every action is logged

The audit trail doesn't distinguish between a human editor and an AI agent. Who changed what, when, through which tool.

Your data stays in your database

The AI Assistant and MCP server operate on your backend. Content isn't copied to a third-party service for processing.

Control Read/Write Access

Decide what your AI agents can read, write, or never touch. Set the rules per field, per collection.

Governance

One permissions system for both humans and AI.

AI agents get a Directus role, just like every other user. Field-level permissions. Collection-level access. The same policy layer that governs your team governs your AI.

Every action is logged

The audit trail doesn't distinguish between a human editor and an AI agent. Who changed what, when, through which tool.

Your data stays in your database

The AI Assistant and MCP server operate on your backend. Content isn't copied to a third-party service for processing.

Control Read/Write Access

Decide what your AI agents can read, write, or never touch. Set the rules per field, per collection.

What this looks like

Real teams. Real workflows.

Three teams, three different jobs. Handled from one backend.

Content Team

See it in action

Translate and publish in one conversation.

A content editor asks the AI Assistant to translate a blog post into French, German, and Spanish. It writes the translations directly into the correct locale fields. No third-party tool, no copy-paste, no manual field mapping. One conversation, done.

Developer

See it in action

Scaffold a data model from your IDE.

A developer connects Cursor to Directus via MCP and asks it to create a new collection with fields, relationships, and validation rules. The schema is live in seconds. No clicking through the Studio, no writing migration scripts.

Operations

See it in action

Tag 500 images before lunch.

An ops manager connects Claude Desktop to Directus and asks it to analyze product images, generate alt text, and apply category tags. The metadata is written directly to the asset records. Hours of manual work, gone.

What this looks like