The newest member of your digital team.
Scout is Uniform's AI agent. It understands your entire project and does real work from a simple request. Everything accomplishment is a draft to review before you publish. Move quickly, ensure quality, stay compliant.
Create a content model and the page that uses it.
Translate this section to German.
Set up an A/B test on the hero.
Example requests, written the way you would ask a colleague.
?width=1600&fit=cover)
Scout will never publish. Every edit submitted as a draft, attributed and reviewable, inside the same workflows, roles, and permissions as the team it is helping.
Scout works exclusively on draft content. Publishing, workflow approvals, and everything live stays behind human decisions by design.
Scout's edits are assigned to the person who asked, visible in history, and bounded by the same roles and permissions as your team.
You choose per task: routine metadata can run autonomous; brand-critical work gets the review queue.
Every task begins with a plain request and ends as a draft for you to review.
Scout creates the composition and the route it lives at. From idea to structured, working content in one conversation. New component definitions, content types with their data types generated so they work end-to-end, block types, entries, and whole pages.
Scout works through the set in a single pass, and Review changes mode lets you inspect every entity before applying anything. This is where a teammate becomes a workforce.
Translate all blog entries to German.
Add an SEO description to every product page.
Retag every image against the new label set.
Use semantic search across compositions, entries, assets, and patterns. If a product is retiring, get a list of all the places the product name appears for a comprehensive change.
Scout sets up personalization and A/B tests on any component, references live performance numbers on demand, and declares winners when the verdict is in.
Scout runs SEO analysis and fixes, generates FAQ content and TL;DR summaries, rewrites copy in citation-friendly style, and refreshes AEO and GEO fields across full sets of pages, helping answer engines cite accurately.
Scout answers questions about the platform, links the correct documents, and helps the team navigate Uniform. The onboarding buddy every new teammate desires.
Four places Scout already lives, so asking is never a detour.
A Scout tab on the project dashboard, with your recent threads.
The sidebar, one click from anywhere in the header.
Inside the component inserter: describe what you want, Scout picks the right component and fills it in.
On any component in the editor: start an AI-assisted personalization or test straight from the highlighter menu.
Scout has context for what you are working on, so a succinct request just works. Stop context-pasting.
Translate this to German.
Skills are your team's reusable instructions written once at the project level. Your brand voice, review checklist, naming conventions, and the way campaign pages are built. Scout loads them automatically or on demand, so the tenth request follows your standards as faithfully as the first.
Attach guidance to specific components and fields too, and Scout respects it across the project.
Example skill names, not product data.
Connect the MCP servers your team uses and Scout will use the tools from inside Uniform. It can pull the ticket, read the spec, and act on content in a connected system.
Uniform's own MCP server lets your external agents and coding tools work with your project too.
Scout, inside your project
MCP runs both ways with your project as the hub. The servers your team runs connect in, and Scout uses them from inside Uniform. Uniform's own MCP server lets your external agents and coding tools connect in too.
Scout Automations removes the repetition. Write instructions the way you would brief a colleague, and Scout will run the instructions every time the trigger fires, with the event as context. Recurring jobs in your content operation become jobs Scout handles.
An entry is saved, a composition is published, and Scout reacts.
when a product entry changes, refresh its SEO fields and OG preview
Nightly, weekly, before every campaign review.
every Monday, refresh the TL;DR fields on the ten most-visited guides
A webhook from your PIM, DAM, or anywhere else.
when the PIM announces a price change, update the affected entries
Developers expose deterministic, permission-scoped actions that Scout can choose to call mid-conversation.
when Scout asks, look up a product's live price and return it
The colleague works nights; the rules do not change.
Automations run with the exact permissions you grant them, consume the same level of AI credits as other Scout work, and provide listed runs that are inspectable.
Automations is in early access, available to select customers today. Talk to us to get on the list
Nothing we've mentioned up to now requires writing code. This is the part that does.
Invoke Scout from server-side code and get structured results back, so agent work slots into your applications and pipelines.
When the job must be deterministic, needs secrets, or calls third-party APIs, write it as code on the same triggers, and call Scout from inside it for the parts that benefit from an agent: deterministic control around an AI step.
Expose your own actions to Scout as typed, permission-scoped tools, running under the identity of whoever invoked Scout.
Straight answers to what teams ask.
Scout is Uniform's AI agent: a teammate inside the platform that knows your whole project (content, models, patterns, releases, optimization) and does real work from a plain request, from creating content models and pages to translating, bulk editing, and setting up tests. Every edit lands as a draft you review.
No, by design. Scout works exclusively with drafts: nothing is published, no workflow advances, and nothing leaves your project until a person acts. Publishing, access, and configuration stay human decisions, which is exactly what makes it safe to let Scout do everything else.
Content models end to end: component definitions, content types with generated data types, and block types. And the content itself: entries, pages with their routes, pattern edits, translations, and assets with generated metadata. Ask for a landing page based on an existing one, and Scout builds it, as a draft.
Three ways. Apply modes per task: Review changes stages every edit for item-by-item inspection, while Autonomous applies drafts directly. Permissions: Scout is bounded by the same roles as your team. And attribution: every Scout edit is assigned to the person who asked and visible in history.
Through Skills (your team's reusable instructions: brand voice, checklists, conventions, loaded automatically) and AI guidance attached to specific components, content types, and fields. Write the way you work once, and Scout follows it on every request, which is how the tenth output matches the first.
Yes, with Scout automations (early access): write instructions once, pick a trigger (a content change, a schedule, or a webhook from another system), and Scout runs them every time it fires, within the permissions you grant. Recurring content work becomes work that simply happens.
Scout runs on AI credits included with Uniform plans, consumed as Scout works, whether in the dashboard or through automations. Talk to us for the allocation that fits your team's volume.
Bring a real task from this week's backlog. Watch Scout do it, and review the result together.
Pick an arduous task from this week: a model that needs building, a section that needs translating, a library that needs retagging. We'll run it with Scout in a project, in Review Changes mode, and go through the pending edits together before applying.
The task, who owns it today, and what a good result looks like.
We make the request in plain language and watch it land as drafts, item by item.
What you would let Scout run autonomously, and what stays in the review queue.
Not ready to talk to anyone? Fair. Read the Scout docs or what an agentic CMS actually is: no meeting required.
.svg){kind=icon}
.jpg)
