Changelog

The dashboard learns what matters most, and agents can read the site

A day of sharpening: the dashboard now leads with the week’s finding, and the whole site became readable to AI agents, which is only fair for a product that measures exactly that.

• The week’s insight (new gaps, with its buttons) moved above the metric cards
• Metric cards come in two sizes now, weighed the way the visibility score weighs them
• Reports, Traffic and Portfolio are reachable on a phone through the More sheet
• Markdown mirrors of the main pages at /index.md, /pricing.md and friends, plus llms.txt, so assistants can quote us accurately
• A Developers footer column, and docs playbooks: Aureo in Claude in one command, and scan results in Slack through a signed webhook relay
• A lapsed subscription keeps every recommendation it paid for, read only
• Report a bug or an idea from inside the app, with steps, a screenshot, and the technical context collected for you
• Checkout now handles EU VAT properly: 25% for Croatia, reverse charge with VAT ID validation for EU businesses, prices stay net

Export, the heatmap, and a plan for people between Starter and Growth

• CSV export of queries, gaps and cited sources
• A per engine heatmap of who each engine names, with the sample size in every column head
• Suggested queries: the generator proposes, nothing enters your set without a click
• The buyer’s journey: discovery, active search and comparison, each with its own reading
• The Plus plan at 119 EUR: 50 queries with recommendations and alerts included
• Starter repriced to 49 EUR with 15 queries; Agency pools 300 queries across 10 brands

Being recommended is not the same as being named

The week the metrics grew up, and the product got a developer surface.

• Soft mentions: an answer that names you without recommending you is counted as what it is
• The visibility score now weighs being recommended above being named
• Citation share: how much of the sourcing behind the answers is yours, per answer and in the report
• Re-ask a single query without moving the week it belongs to
• A REST API with scoped keys, an MCP server for Claude and Cursor, and signed outbound webhooks
• The /sources screen: every domain the engines cite in your niche, by format

Aureo reads the site it advises on

• Every scan now checks your own site first: robots.txt, the AI crawlers it admits, llms.txt, and whether GPTBot actually gets a page
• Recommendations argue from what your site already publishes, so "write this page" becomes "fix this URL" where that is the truth
• Confidence intervals on the headline metrics: a move inside the sample noise is shown grey, not claimed
• Agency logo upload for white label reports
• Settings is one screen again, and every screen shows its shape while it loads

Queries get their own screen

• Add and retire queries on the queries screen, with the pooled plan limit enforced where you type
• Competitor suggestions from the names the engines keep bringing up
• Scheduled mail arrives at eight in your own morning, wherever you are
• One company counted once per answer, with its domain worked out from its own citations

The first version

• Weekly scanning across ChatGPT, Perplexity and Gemini
• Share of voice, mention rate, average position and citation presence
• Citation gap analysis with the sources each engine quotes
• Recommendations with the evidence they were derived from
• PDF report and weekly email