strataDocs
Sign inGet started

AI Assistant

Strata's AI Assistant helps you generate content types, review your architecture, ask questions about your content model, summarize recent changes, and plan migrations — all powered by your own API key.

Overview

The AI Assistant is a bring-your-own-key (BYOK) feature. Strata does not provide AI credits or proxy your requests through its own infrastructure. Instead, your organization connects its own API key from a supported provider, and all AI usage is billed directly to that key. Your API key is encrypted at rest and never shared.

AI features are available on the Business and Enterprise plans. Free and Team plans do not include AI access. If your organization is on a lower tier, the AI tab will not appear and the AI settings page will redirect to your general settings. See Billing for details on plan tiers.

Supported providers

Strata supports a range of AI providers. You can choose from preset models or enter a custom model ID for any provider:

  • Anthropic — Claude Sonnet 4, Claude Haiku 4, Claude Opus 4
  • OpenAI — GPT-4o, GPT-4o Mini, o3-mini, o4-mini
  • Google — Gemini 2.5 Pro, Gemini 2.5 Flash
  • Mistral, Groq, Together AI, Fireworks AI — popular inference providers with preset models
  • Custom (OpenAI-compatible) — any provider that exposes an OpenAI-compatible API endpoint

Setting up your API key

Only organization owners can configure AI settings. To connect a provider:

  1. Go to Organization Settings and select the AI tab.
  2. Choose a provider from the dropdown (e.g. Anthropic, OpenAI, Google).
  3. Paste your API key. The key is encrypted before storage and displayed as a masked value after saving.
  4. Select a model from the preset list, or choose “Custom model ID” to enter one manually.
  5. Click Save configuration.

After saving, use the Test connection button to verify that the key and model are working. If the test fails, check that your key is valid and has sufficient permissions.

For custom OpenAI-compatible providers, you will also need to provide a Base URL — the API endpoint your provider exposes (e.g. https://api.example.com/v1).

You can change the provider or model at any time by returning to the AI settings page. To remove the configuration entirely, click the delete button next to the current configuration.

The AI hub

Once configured, the AI tab appears in the project sidebar. Clicking it opens the AI hub, which has five feature tabs:

  • Generate — create content types and fields from a natural-language description
  • Review — run an automated architecture review of your content model
  • Chat — ask questions about your project's content model
  • Summarize — generate a narrative summary of recent changes
  • Migration — analyze pending changes and produce a step-by-step migration plan

The hub header shows which provider and model are currently active. Each tab maintains its own session history, so you can save, reload, and continue previous sessions.

Generate

The Generate tab lets you describe the content architecture you need in plain English. The AI reads your project's registry (available field types, widgets, and formatters) and returns a structured proposal with content types, taxonomy vocabularies, media types, and their fields.

How it works

  1. Type a description of what you need, for example: “I need a blog with categories, tags, featured images, and author profiles”.
  2. The AI generates a proposed content model — a set of bundles (content types, vocabularies, etc.) each with their fields, machine names, cardinality, and field types.
  3. Review the proposal. Each bundle is shown as an expandable card with its full field table (label, machine name, type, required, cardinality).
  4. Select the bundles you want to keep using the checkboxes, then click Apply Selected to add them to your project.
  5. Continue the conversation to refine the model — ask for changes, add more content types, or adjust fields.

Contrib module suggestions

If the AI proposes field types that are not in your project's registry (e.g. smart_date from the Smart Date module), they are flagged with a warning. The proposal will also list suggested modules with links to drupal.org, so you know which contributed modules to install to support those field types.

Cross-project context

Toggle Include other projects as context at the bottom of the Generate tab to let the AI see the content models from your other projects in the same organization. This helps the AI suggest consistent naming and avoid duplicate structures.

Review

The Review tab runs an AI-powered audit of your existing content model. Click Run Review and the AI will analyze all of your content types, fields, vocabularies, and relationships.

The review returns:

  • Strengths — things your content model does well
  • Findings — issues categorized by severity (error, warning, or suggestion), each with a description, the affected entries, and a recommendation

Common findings include naming inconsistencies, missing description fields, unused vocabularies, overly broad entity references, and structural improvements. Expand any finding to see its full detail and recommendation.

You can re-run the review at any time after making changes to see updated results.

Chat

The Chat tab gives you a conversational interface to ask questions about your project's content model. The AI has full context about all content types, fields, vocabularies, and their relationships.

Example questions:

  • Which content types reference the Tags vocabulary?
  • What fields are shared across multiple content types?
  • Show me all media reference fields in the project
  • Which content types don't have a description field?
  • Summarize the overall content architecture

Like Generate, you can toggle Include other projects as context to let the AI compare across projects in your organization.

Summarize

The Summarize tab generates a narrative summary of recent changes to your content model, drawn from committed changesets. This is useful for sprint reviews, client updates, or internal documentation.

Before generating, choose:

  • Time range — last 7, 14, or 30 days, or all time
  • Tone — Stakeholder (non-technical, focused on outcomes) or Technical (detailed, focused on field-level changes)

The summary includes change statistics (created, updated, deleted) and a list of key changes. Click Copy to copy the summary to your clipboard for pasting into emails, Slack, or documents.

Migration

The Migration tab analyzes your latest changeset and generates a step-by-step migration plan for implementing those changes in a live Drupal site. Click Analyze Latest Changes to run the analysis.

The migration plan includes:

  • Overall complexity — trivial, simple, moderate, or complex
  • Ordered steps — each categorized as a safe change, config change, data migration, module dependency, or potential data loss risk
  • Risk level per step — none, low, medium, or high
  • Commands — suggested Drush or shell commands for each step where applicable
  • Warnings — highlighted risks such as potential data loss from field removals or type changes

Click Copy to copy the full plan as text. You can re-run the analysis after committing additional changes to get an updated plan.

Session history

Each AI feature tab maintains its own session history. You can save the current session, start a new one, or reload a previous session from the session bar at the top of each tab. Saved sessions persist per project, so your team can refer back to earlier AI interactions.

Next steps