> ## Documentation Index
> Fetch the complete documentation index at: https://docs.flowx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Architect

> Generate and extend project data models from natural language prompts, with a preview step before you apply anything to your project.

export const release_0 = "5.11"

## Overview

The AI Architect is a design-time agent that turns natural language descriptions into project data models. Describe the business entities you need, with their attributes, types, and relationships, and the AI Architect proposes a data model you can preview in the data model builder and apply to your project. Follow-up messages refine or extend the proposed model instead of starting over, so you can build up a model iteratively.

<Info>
  **Available starting with FlowX.AI 5.9.2.** The AI Architect is not available in 5.9.0 or 5.9.1 — it requires the 5.9.2 platform and AI Platform components.
</Info>

***

## Key capabilities

<CardGroup cols={2}>
  <Card title="Generate a data model from text" icon="keyboard">
    Describes-to-model: interprets a natural language description of your business domain and proposes entities, attributes, and types for your project data model.
  </Card>

  <Card title="Extend an existing model" icon="diagram-next">
    Follow-up messages continue the same conversation and refine the current proposal without regenerating from scratch. Against a model already saved to the project, the agent only adds new entities and attributes — it does not rename, retype, or remove existing data types.
  </Card>

  <Card title="Preview before applying" icon="eye">
    The proposed model renders as a preview in the data model builder graph, so you can inspect entities and attributes before anything is written to the project.
  </Card>

  <Card title="Apply to your project" icon="check">
    Applying the preview creates the new entities and attributes in your project data model in one operation. Apply only adds — existing data types are never modified or removed. If a proposed type name collides with one already in the project, it is created under a new name and the conversation lists which types were renamed.
  </Card>
</CardGroup>

***

## How to use

<Steps>
  <Step title="Open the data model builder">
    In your project, open the **Data Model** section.
  </Step>

  <Step title="Generate with AI">
    Select **Generate with AI** and describe the data model you need — the entities, their attributes, and how they relate.&#x20;
  </Step>

  <Step title="Review the preview">
    The proposed model appears as a preview in the data model graph. Inspect the entities and attributes before accepting anything.
  </Step>

  <Step title="Refine with follow-ups">
    Send follow-up messages to extend or adjust the proposal. The agent continues the same thread instead of generating a new model.
  </Step>

  <Step title="Apply the model">
    When the preview matches what you need, apply it. The new entities and attributes are created in your project data model; existing data types stay untouched.&#x20;
  </Step>
</Steps>

***

## Generate a data model from a document

<Badge color="blue" icon="cloud">SaaS · {release_0}</Badge>

<Info>
  **Available on SaaS with FlowX.AI {release_0}.** This feature is live on managed (SaaS) deployments now. Self-hosted deployments will receive it with the next LTS release family.
</Info>

Instead of describing the model in text, you can have the AI Architect build it from an existing document — for example, a data dictionary, a specification, or a form. The agent reads the document, extracts the entities and attributes it describes, and proposes a project data model that you preview and apply the same way as a text-generated model.

<Steps>
  <Step title="Open Generate with AI">
    In your project's **Data Model** section, select **Generate with AI**.
  </Step>

  <Step title="Choose From document">
    In the initial message, select **From document** and pick the document to generate from. The agent extracts the relevant entities and attributes from the document's content.
  </Step>

  <Step title="Review and edit the proposal">
    Inspect the proposed model in the preview. Before applying, you can adjust it — add keys by prompting the agent in a follow-up message, or use the **Modify keys** modal to edit keys directly.&#x20;
  </Step>

  <Step title="Apply the model">
    Apply the preview to create the new entities and attributes in your project data model. Apply only adds — existing data types are never modified or removed.&#x20;
  </Step>
</Steps>

***

## Best practices

<Card icon="square-check" color="green" horizontal="true">
  **Describe entities and their relationships:**
  Name the business objects you need and how they relate. For example, "`a Customer with name, email, and date of birth, and a list of Accounts, each with an IBAN and a balance`".
</Card>

<Card icon="square-check" color="green" horizontal="true">
  **Refine instead of regenerating:**
  Use follow-up messages to add or adjust entities in the proposed model. The agent extends the existing proposal, which keeps the preview consistent.
</Card>

<Card icon="ban" color="red" horizontal="true">
  **Don't apply without reviewing the preview:**
  The preview exists so you can check attribute names and types against your conventions before they are created in the project.
</Card>

***

## Related resources

<CardGroup cols={2}>
  <Card title="AI Designer" icon="robot" href="./ai-designer">
    Generate UI layouts that map to your data model
  </Card>

  <Card title="AI Developer" icon="code" href="./ai-developer">
    Generate business rules and integration logic
  </Card>

  <Card title="Agent Builder" icon="hammer" href="../agent-builder/overview">
    Overview of AI agents and the Agent Builder
  </Card>
</CardGroup>


## Related topics

- [FlowX.AI architecture](/5.9/docs/platform-overview/flowx-architecture.md)
- [AI Platform setup](/5.9/setup-guides/ai-platform-setup.md)
- [AI Gateway setup](/5.9/setup-guides/ai-gateway-setup.md)
- [Deployment guidelines v5.11](/release-notes/v5.x/v5.11.0-august-2026/deployment-guidelines-v5.11.md)
- [Deployment guidelines v5.9.2](/release-notes/v5.x/v5.9.2-july-2026/deployment-guidelines-v5.9.2.md)