Building experiences with AI

Uniform's AI tools help teams build digital experiences faster while keeping content and architecture consistent. Content creators and marketers use Scout and AI quick edits directly in the dashboard. Developers and experience architects use the Uniform MCP server from their AI coding assistant, author skills so Scout follows their team's procedures, and connect external MCP servers so Scout can reach third-party systems. All of these benefit from AI guidance, which teaches the AI your brand voice, content requirements, and architectural standards.

note

All AI features require AI credits. Ensure your team has credits allocated before getting started.

Understanding when to reach for each tool helps you work more efficiently.

  • Multi-component and multi-entity edits - Make changes across multiple components or entities in one conversation, queued as pending edits for review when needed
  • Semantic search across the project - Find compositions, entries, assets, and patterns by meaning, e.g. "entries about coffee machines"
  • Release-aware work - Search, list, inspect, and edit content inside releases
  • End-to-end optimization setup - Create simple query-string signals and do full CRUD on quirks, intents, audiences, and enrichments; configure personalization and A/B tests
  • Conversational workflows - Iterate with back-and-forth refinement, and learn Uniform along the way

Example prompts:

  • "Create a new composition named 'Product Launch 2026' with a Hero component, Features component, Testimonials component, and CTA component"
  • "Create a new Article content type with title, featured image, author reference, and rich-text body"
  • "Find every 'Product' entry about coffee machines and check the 'Sustainable' field on the ones that mention recycled materials"
  • "Translate this landing page into French and German"
  • "Add personalization to the hero component targeting the 'First-Time Visitors' and 'Returning Customers' audiences"

For the full list of what Scout can and can't do today, see Scout capabilities and limitations.

  • Single-field updates - Fast content generation for individual parameters or fields
  • Translation - Quickly translate content to other locales
  • Content refinement - Rephrase, shorten, or improve existing text
  • A/B test variations - Generate alternative copy for testing
  • Field-specific content - Leverage field-level AI guidance for targeted content

Example prompts:

  • "Translate this to German"
  • "Make this headline more concise, under 60 characters"
  • "Rewrite this in a professional tone suitable for enterprise audiences"

Learn more in the AI quick edits documentation.

Use the Uniform MCP server when you want to manage Uniform from your IDE alongside the code that renders it.

  • Code-adjacent workflows - Author components, content types, and definitions next to the code that renders them
  • Bulk and scripted schema changes - Manage many definitions efficiently from a single conversation
  • Version control integration - Keep your local Uniform project files in sync with the dashboard and commit changes
  • Multi-MCP setups - Combine with other MCP servers (Figma, Context7, Storybook) for end-to-end design-to-code flows

Example prompts (in your AI coding assistant):

  • "Create a Card component with parameters: image (asset type), title (text), description (rich text), and buttonText (text)"
  • "Add a new 'author' field (text type) to the Article content type"
  • "Pull the latest content types from Uniform and sync them to my local project"

Learn more in the Uniform MCP server documentation.

Use skills when your team has procedures Scout should repeat the same way every time, such as naming conventions, SEO checklists, or the steps you take whenever you create a landing page.

  • Author once, apply everywhere - Every conversation in the project gets the same procedure
  • Always-included for guardrails (brand voice, naming, "always do X"); on-demand for procedure skills Scout loads when a request matches

Connect external MCP servers when you want Scout to reach out of Uniform, for example to fetch a Linear issue, pull some draft content from Notion, update a GitHub issue, or query a third party CMS or PIM system.

  • Set up per-tool enable/disable so Scout only calls what you need
  • Pair with the Review changes apply mode the first few times for review-first runs
  • Review the security and trust callout before connecting a new server

Building a new landing page#

Tools: Scout + AI quick edits

  1. Open Scout and describe the page you want to create
  2. Scout creates the composition structure with components
  3. Use AI quick edits to refine individual headlines, descriptions, and CTAs
  4. Optionally, set up personalization or A/B tests (see Optimize with AI)
  5. Review and publish

Example: "Create a new composition named 'Analytics Product Launch' with these components: Hero component with headline and CTA, Benefits component with three feature cards, Testimonials component, and Pricing component."

Localizing content#

Tools: Scout or AI quick edits (depending on scope)

For complete pages:

  1. Open the composition in the target locale
  2. Ask Scout: "Translate all content on this page to French (fr-FR locale)"
  3. Review translations for accuracy and cultural appropriateness

For individual fields:

  1. Navigate to the field in the target locale
  2. Click the AI quick edit icon
  3. Use prompt: "Translate this to Spanish (es-ES locale)"

Setting up a new content model#

Tools: Scout (in dashboard) or MCP server (in your IDE)

Pick the tool that matches where you live day-to-day:

  • In the dashboard: ask Scout to create the component and content type definitions. Useful when content modelers and editors are the same people.
  • In your IDE: use the MCP server so changes show up next to the code that renders them, then sync and commit.

Example: "Create a BlogPost component with these parameters: heroImage (asset type), headline (text), authorName (text), publishDate (date), body (rich text), and relatedArticles (content reference to Article content type, multi-select). Also create an Author content type with fields: name (text), bio (rich text), and photo (asset)."

If you used the MCP server route, review in Uniform and commit to version control using the sync command.

Scaffold computed-like fields with quick edits + field guidance

For fields that should be derived from other fields on the same entity (an SEO description from the body, a short summary from a long one, an OG title from a headline), set field-level AI guidance describing how the field should be generated, then trigger an AI quick edit with no prompt. The guidance alone produces a consistent default, no chat required — a lightweight stand-in for computed fields.

Scout can analyze pages, fix SEO issues, set up AEO and GEO basics, and generate social previews. For A/B testing and personalization workflows that optimize conversions, see Optimize with AI.

Scout can analyze pages and provide SEO recommendations, then automatically fix common issues:

Running an SEO audit: Ask Scout: "Run a comprehensive SEO audit on this page and automatically fix all critical issues" to identify and resolve issues with meta tags, headings, content structure, and more.

Fixing individual fields: Use AI quick edits for targeted optimization of meta descriptions, title tags, and alt text.

Example prompts:

  • "Populate the seoDescription field on this product composition, emphasizing the keywords 'analytics platform' and 'data insights'."
  • "Generate the altText field on the hero image asset used by this product composition."

SEO best practices: Configure AI guidance on SEO fields to ensure consistent optimization. For example, set guidance on meta description fields to always include primary keywords and stay within 150-160 characters.

AEO (answer engine optimization) and GEO (generative engine optimization) reward structured, attributable, citation-friendly content over polished prose. Uniform's composable content model produces exactly that, so the most effective approach treats AEO and GEO as a content modeling problem rather than a prompting problem.

AEO vs GEO

Answer engine optimization (AEO) structures content so engines like Google AI Overviews, Bing Copilot, Perplexity, and ChatGPT search can extract direct answers: FAQ sections, question-form headings, and structured data. Generative engine optimization (GEO) makes content quotable and citable inside the generative summaries those engines produce: TL;DR fields, concrete data points, author attribution, and clear publication dates.

Three steps to set up AEO and GEO:

  1. Model the AEO/GEO building blocks once. Add fields like tldr (rich text), author (reference), datePublished, dateModified, and a short description (text) to the content types where they belong. For structured Q&A content, create a separate FAQ content type with question and answer fields and link to it from the parent with a reference field. This pattern avoids relying on field shapes Scout can't edit. Scout can create and update content type definitions, so the modeling work itself can be a Scout prompt.
  2. Set field-level AI guidance describing what good looks like on each field (length, voice, citation style, question form). The standard then applies automatically to every Scout edit, AI quick edit, and MCP edit.
  3. Run a Scout pass per entity for iteration, or as a batch edit for site-wide refreshes. Switch to Review changes mode for publish-bound runs.

Running an AEO/GEO refresh: Ask Scout: "Run a project-wide AEO and GEO refresh across every product page: link FAQ entries to the page, populate the TL;DR, and set the author attribution. Use Review changes mode." Scout works through the set in a single pass and queues each entity for review.

Fixing individual fields: Use AI quick edits for targeted updates to tldr, FAQ answers, and structured-data fields. With field-level guidance in place, you can often run a quick edit with no prompt and get a consistent default.

Example prompts:

  • "Add a tldr rich-text field to the BlogPost content type. Create a separate FAQ content type with question (text) and answer (rich text) fields, then add an faqs reference field on BlogPost so each post can link the FAQs that belong to it. Also create an Author content type with name, bio, credentials, and photo, and add an author reference field to BlogPost."
  • "Create eight FAQ entries that answer the most common questions a first-time buyer would ask about this product, and reference them in the product entry's faqs field."
  • "Rewrite the headline parameter on this composition's Hero component, and the title parameter on each section component below it, as questions a user would type into an AI assistant."
  • "Generate the tldr field on this blog entry from its body content."
  • "Populate the structured-data fields on every BlogPost entry: headline, datePublished, dateModified, author, and seoDescription."
  • "Refresh the body rich-text field on this blog entry for AEO and GEO."

AI crawlers see your default variant

AI crawlers (Google AI Overviews, Perplexity, OpenAI, and others) typically don't trigger personalization signals, so they see whatever the default variant renders. Make sure the default variant carries the AEO/GEO content (TL;DR, linked FAQ entries, structured-data fields), even when richer personalized variants exist. When declaring a winner on an A/B test, pick the AEO/GEO-friendly variant as the default whenever the metrics are close.

AEO and GEO best practices: Set AI guidance on tldr and structured-data fields, and on the question and answer fields of the FAQ content type, so every new entry gets a sensible default. Model the structured-data fields once and let Scout populate them rather than asking it to invent JSON-LD markup. For your team's recurring AEO/GEO checklist, author two skills: one always-included for guardrails, and one on-demand for the refresh procedure.

Use Scout to generate open graph tags and social media previews:

Example prompts:

  • "Populate the Open Graph fields — ogTitle, ogDescription, and ogImage — on this product composition."
  • "Refresh the ogTitle and ogDescription fields on every case study composition."
  • "Set the twitterDescription field on this blog entry about AI in marketing."

For content creators#

  1. Set up AI guidance at the project level for your brand voice
  2. Start with Scout to build your first AI-assisted composition
  3. Explore AI quick edits for efficient field-level updates
  4. Learn about conversion optimization with AI for A/B testing and personalization
  1. Install and configure the MCP server
  2. Create your first component using natural language
  3. Set up AI guidance for your components and content types
  4. Author skills so Scout follows your team's procedures inside the dashboard
  5. Connect any external MCP servers your editors need (Linear, Notion, GitHub, …)
  6. Integrate Uniform sync into your development workflow

info

AI is a powerful assistant, but always review and refine AI-generated content to ensure it meets your specific requirements and brand standards.

  • Optimize with AI - Set up A/B testing and personalization for conversion optimization
  • Scout - Detailed guide on Uniform's AI agent capabilities
  • AI quick edits - Fast single-field content generation and refinement
  • AI guidance - Configure AI to understand your brand voice and content standards
  • Skills - Teach Scout your team's repeatable procedures
  • MCP server - Connect Uniform to your AI coding assistant
  • Connect external MCP servers - Let Scout reach third-party tools
  • AI credits - Manage your team's AI usage
Last modified