Every change, dated.

Tracking now runs every 24 hours by default on Growth and Scale, no scheduling required. Engines that previously ran on a 72-hour cycle (DeepSeek, Mistral, Llama-Index) are caught up to the same cadence. The dashboard surfaces a "last refreshed" timestamp on each prompt so you always know how fresh the citation count is.

Approve a schema or content fix in the Clovion dashboard and it lands as a draft post (or page update) in your connected WordPress site. The connector preserves your editor of choice — Gutenberg, Classic, or Elementor — and never publishes without a human approval step.

Tracked separately from Claude 3.5 so you can watch the migration as it happens. Both versions are scored on the same prompt set with identical sentiment models; the comparison view in your dashboard now defaults to showing them side by side.

Drop a CSV with up to 5,000 prompts and Clovion handles deduping, locale detection, and intent classification automatically. Useful if you have years of search query logs you want to graduate from keyword research to prompt research.

A maintenance week focused on the dashboard and the suggestion queue. Nothing flashy, but enough quality-of-life polish that the team felt it deserved its own entry rather than burying it inside the next release.

The Clovion AI MCP server is public beta. Point Claude or ChatGPT at your workspace and they can query your visibility data, pull the suggestion queue, and ship approved fixes — all from inside the assistant you already use. Authentication uses scoped API tokens and respects the same RBAC as the dashboard.

A citation in the first sentence of a Perplexity answer is worth more than a citation in the footnote list. Our previous model treated them equally; the new one weights position, paraphrase depth, and answer prominence. Existing scores have been recomputed and historical charts updated.

Our Type II audit covering October 2025 through September 2026 wrapped with zero exceptions. The report is available under NDA from the Trust Center; ISO 27001 is mid-audit and tracking for Q1 2027. Nothing in the product changes — this is paperwork that lets larger buyers move faster.

Mid-month polish before the bigger MCP work landed. Mostly papercuts surfaced by the customer team — the kind of thing one person hits once and never reports, but seven people hit and quietly route around.

A new endpoint fires when a suggestion is created, approved, shipped, or measured. Wire it to your build system, CMS, or Slack and the loop closes without anyone tabbing into the dashboard. Signatures are HMAC-SHA256 and replay-safe with a 5-minute window.