The AI Visibility report now has a "Visibility over time" chart. It shows mention rate broken down by topic, prompt, persona, and platform — so you can see whether your position is actually moving, not just what it is today. Scans run close together are pooled into a single data point so re-scans don't read as false movement, and low-sample lines fade to signal where the data is thin.
The visibility score works differently now. Answers where the AI wrote prose without recommending any vendor at all — no winner, no loser — no longer count as full misses against you. Answers that named a competitor still do. The headline score also weights core recommendation prompts more heavily than broad aspirational ones, so adding ambitious prompts to your set no longer mechanically pulls the number down. Every surface that shows a visibility score — report, dashboard, brand index, investor view — now draws from the same formula, so they can't contradict each other.
Scan depth has increased. The expanded query cap for pro-tier scans is now more than double what it was, giving a typical prompt set real persona-level depth instead of mostly baseline-only coverage. When a scan's query budget is tight, core recommendation prompts get depth allocated first — the prompts where you're actually competing for a slot. The per-scan cost stays near the same target because the cheaper model set is now the default across all tiers, with premium models still available if you want them.
If you signed up after typing a URL into the content engineering preview, onboarding now pre-fills that URL automatically. You no longer have to re-enter what you already typed before checkout.
Brand URLs without a protocol prefix (like Pendium.ai) now resolve correctly as outbound links on agent sites instead of routing internally.