Agents With Guardrails
AI agents in RIFT never touch the database. Every agent action runs through governed, logged tools and a human-review fallback.
RIFT is a cutting edge CMS built for your most technical and precise publishing flows.
AI agents in RIFT never touch the database. Every agent action runs through governed, logged tools and a human-review fallback.
Drafting agents read your existing pages before assisting with content authoring. New pages match your voice, apply your design system components, and cross-linking to content you've already published instead of inventing orphan pages.
Write your own markup, styles, and scripts. RIFT doesn't sanitize your expertise away — it versions it, audits it, and stages it like everything else.
Content that already exists as JSON, HTML, or XML enters RIFT as structured content.
Your design system lives inside RIFT — versioned, documented, and enforced at authoring time for humans and agents alike. Derive a design system automatically from an existing site, or build one from scratch within RIFT.
RIFT parses links on every save and maintains a real-time content graph. Before you retire, move, or rename anything, you see the blast radius. Impact analysis is a query, not an archaeology project.
A four-state workflow with paired staging means every change is previewable in a production-faithful environment before it ships.
Every content item carries full version history. Policy pages, disclosures, help content — you can show exactly what was published, when, and by whom, for the entire life of the page.
Optimistic locking on every edit means two editors can't quietly destroy each other's work. Conflicts surface immediately, at save time — not as a surprise in next week's review.
Accessibility scanning runs as part of the platform, with audit reports — not as a third-party plugin bolted on after the complaint arrives.
Every publish is a versioned commit to your GitHub repository, deployed to Cloudflare, Netlify, or Vercel — whichever you already trust. No proprietary hosting, no lock-in, and a complete audit trail for free.
RIFT publishes static files. Your live site doesn't run on RIFT — it's already deployed. A CMS outage, upgrade, or migration never takes your public site offline.
Every page publishes as a website for people and an Open Knowledge Format bundle for AI agents — structured, machine-readable, and ready for the way agents actually consume the web. No scraping layer, no custom RAG pipeline.
Your app teams can use RIFT API calls and MCP. Every publish, edit, and rollback runs through a versioned API with audit logging, idempotency keys, and dry-run mode. Automate anything; trace everything.
RIFT's migration agents crawl your legacy site, detect its shared chrome, classify every page by content type, and discover the repeating design patterns hiding in your markup.
The pipeline proposes a formal design system derived from your existing pages, then parameterizes your content against it — so what lands in RIFT is structured, not screenshots of the past.
Ambiguities in migration pause the pipeline and route to a human for design decisions, edge-case pages, judgment calls. You approve the choices that matter; the agents handle the repetition.
The full pipeline has run end to end on a 500+ page federal help site: chrome detected, design system extracted, content classified and landed, accessibility preserved.