RIFT is built for teams whose publishing decisions get audited — human approval on every AI action, a complete version record, and infrastructure that answers hard questions before you're asked. See the full /featuresfeature set→ or request access below.
Every AI agent action in RIFT — drafting a page, editing a component, proposing a link — produces a change-set, not a live edit. A named person on your team reviews, edits, or rejects it before it reaches the public site.
Every content item carries full version history — what changed, when, and who approved it. Policy pages and disclosures can show their entire publication history on demand.
Accessibility scanning runs as part of the platform itself, with audit reports generated automatically — not a third-party plugin bolted on after a complaint arrives.
There is no agent path that bypasses review. Every draft, edit, or proposed link an AI agent makes is staged as a change-set and routed to a named human before it can touch the live site — approve, adjust, or reject, every time. Learn more →
Every content item retains its complete version history for the life of the page. When an auditor asks what was published on a disclosure page and when it changed, the answer is a lookup — not a reconstruction project. Learn more →
Publishing pushes a versioned commit to your own GitHub repository, deployed to Cloudflare, Netlify, or Vercel — whichever you already trust. No proprietary hosting, no lock-in.
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 publish, edit, and rollback runs through a versioned API with audit logging and idempotency keys — trace any change end-to-end, months or years later. Learn more →
RIFT's AI routing is configured per instance, not fixed in the platform. On self-hosted government deployments, no content or prompts leave your environment to an external provider unless you explicitly wire one in. Learn more →
The versioned API and MCP tools your agents use route entirely within your deployment boundary — configure providers, review call logs, and audit every request without traffic crossing into a third-party service. Learn more →