Changelog

Most AI visibility data tells you whether your brand appears in AI responses. Agent Analytics goes one layer deeper: it shows you what AI crawlers are actually doing on your site, which pages they visit, how often, and whether they can access them at all. Built on your own server log data, it gives a direct and objective view of how AI bots interact with your content. That helps you catch technical access issues before they affect your visibility and understand which pages AI models are most interested in.

At launch, Agent Analytics includes two tools: Crawl Insights and Crawlability. Log ingestion connects via Cloudflare Workers or CSV/CLF file upload. Support for additional hosting providers is on the roadmap.

Crawl Insights

Connect your server logs and see which AI bots visit your site, which pages they hit, how often, and with what result. The overview shows aggregate trends across bots, status codes, and traffic volume. The details view lets you drill into specific pages, filter by bot type or time range, and see how retrieval data from your tracked prompts sits alongside your actual crawl activity. That combination is directional and useful for finding where to investigate, not a direct cause-and-effect relationship.

Crawlability

Check which AI bots are allowed or blocked by any domain's robots.txt instantly, without needing server access. You can also test individual URLs against the current rules to catch accidental blocks before they affect how AI models reach your content.

You can now use your Peec AI data directly inside the AI tools you already work in with the new Peec AI MCP. Ask Claude, Cursor, VS Code, Windsurf, and other MCP-compatible tools questions about your visibility, competitors, sources, prompts, and trends using your real Peec project data.

This makes Peec AI more than a dashboard. Instead of switching between tools, exporting reports, or pulling screenshots by hand, you can get answers where the work is already happening. That makes it easier to build reporting workflows, run competitive analysis, and turn AI visibility data into action faster. To make setup easier, you can now find Peec AI MCP directly in partner connector directories (Anthropic & Cursor) and connect with a single click.

More of your data is now available inside MCP

The MCP server now gives you deeper context when you ask questions. You can pull source URL content for analysis, see YouTube channel names and Reddit community names in source reports, access Peec AI recommendations directly through MCP, and inspect available AI models without guessing valid model IDs.

MCP now supports more complete workflows

You can now go beyond read-only analysis. We shipped native prompt workflows for faster repeated tasks, added write tools so you can create, update, and delete key Peec AI entities from MCP clients, and improved the server instructions so AI tools are better at choosing workflows and recovering from errors on their own.

Your topics and prompts are only as good as what Peec knows about your brand. Brand Profile gives you one place to describe your brand, market, and target audience. When you update it, Peec automatically regenerates your suggested topics and prompts so your tracking reflects how your brand actually shows up in AI responses, not a generic approximation.

We also redesigned the Brands page. It now uses a table layout with inline editing for display names and brand colors, and a cleaner flow for managing domains. Less friction across the board when you're setting up or updating your competitive landscape.

Pause and resume projects

You can now pause any tracking project and resume it whenever you are ready. Your setup stays intact, and credits stop being consumed while paused. Useful if you are winding down a campaign, handing off to a client, or want to free up credits without losing your configuration.

New brand filter

The brand filter is now multi-select and grouped into "Your brand" and "All brands" with search. Keep your own brand as the anchor point and choose exactly which competitors to compare against. The Top 7 Brands card on the overview updates with your selection.

Custom domain selection on source charts

You can now pick up to 5 specific domains to display on the source chart. Colors match your brand configuration. Useful if you want to focus on the domains that matter most instead of seeing the default top results.

You can now click into any URL from the Sources page to see the full picture: retrievals over time, citation rate, retrievals by model, which prompts retrieve it, which brands appear alongside it, and the actual chats. All with the same filtering you are used to (date range, model, country, topic).

This was one of our most requested features. Until now, you could only see aggregated data at the domain level. You'd know "reddit.com cites you 50 times" but couldn't see which specific threads drive those citations or how individual URLs trend over time. The URL detail page closes that gap.

Source metrics restructuring

We restructured the metrics in the Sources tables to make them clearer. The old "Used" column was confusing because values could exceed 100%. The new metrics form a funnel: reach, depth, impact.

Retrieved shows the share of chats where a source appeared, always 0 to 100%. Retrieval Rate shows the average number of source appearances per chat as a decimal. Citation Rate shows how often a retrieved source gets explicitly cited.

We also added Citation Rate to the gap analysis view.

Separate Domains and URLs pages

We split the Sources section into dedicated Domains and URLs pages in the sidebar. Previously, URLs were hidden behind a tab that most users never found. Now each view has its own entry point, making it much easier to navigate.

New model selector

Models are now grouped into Active, Inactive, and Available categories. Inactive means deprecated or disabled models that still have historical data. The available models list also shows which models use scraper-based tracking vs API-based tracking.