Genezio Changelog

Perception Monitoring

Pick a claim about your brand and Genezio tracks it across every run: how often AI answers agree or disagree, how that agreement moves over time, and how it differs from one model to the next. Each verdict keeps the pages it cited, so you can see which domains sit behind the answers that get you right and the ones that get you wrong.

Public API and API keys

Account owners create API keys, each limited to the brands it needs, and read Genezio data from their own reporting stack. The API covers visibility, recommendation, citations, sentiment, fact checking, query fan-outs, products and conversations, and manages topics, scenarios, competitors and personas. It is described in OpenAPI and documented with a full reference.

ChatGPT Ads

ChatGPT now shows sponsored placements inside answers. A new Ads page captures each ad in your tracked conversations and shows who advertises on your topics, which creatives they run, the topics and scenarios that trigger them, and your share of ad visibility against competitors.

Agent Analytics

Upload an access log from your site or CDN and see which AI bots visit, what they read, how your server answers them, and how often the pages they read end up cited in AI answers. Each visit is checked against the IP ranges that the AI providers publish, so a fake user agent shows up as unverified instead of inflating the count.

Crawlability

For each website of a brand, Crawlability reads robots.txt and the sitemaps and shows which of 49 AI bots can crawl the site. A host is flagged when a main AI search bot is blocked, when robots.txt is unreachable, or when no valid sitemap exists.

Custom roles

Account owners define their own roles and give each one a level for every area of the product. Members and invites take a custom role or a built-in one.

Content Hub: plans, formats and video

Content Hub now shows the plan of an article for your approval before it writes, and one run produces every format you ask for.

• Brand skills: house style and structure rules, written once or uploaded from PDF, Word or Markdown files, that the agents follow
• Content Analysis reads a YouTube link and gives a video analysis, with a partial report when the video has no transcript
• Topics in New Content have search and filters

Scorecard goals over MCP

An MCP client such as Claude can now create and change Scorecard goals. The goal agent does the build, the same way it does in the dashboard.

Scorecard

Describe a goal in a conversation, for example the share of recommendations you win against one competitor in one market. The goal agent writes the metric, runs it on yesterday's data every day, and puts it on a board you arrange yourself.

• Views for checklists, step lines, share bars and day grids
• A row that names a product shows the product's photo
• Every action on the board goes to the audit log

Content Hub editing

The New Content wizard starts with the content type, and a brief can generate its own type. Each version highlights the text that changed, and an article gives its schema.org JSON-LD markup.

Content Analysis for landing pages and lists of URLs

Content Analysis has a mode for landing pages, and it can start one analysis for each URL of a list.

Topics from a Search Console export

A brand can seed its topics from a Search Console CSV export, which works without a live connection.

Personas with a country

Each persona shows its country, and its conversations take that country's language.

Google Analytics 4 integration

A native GA4 connection that correlates AI visibility and recommendation metrics with downstream web traffic behaviour, including attribution for AI-influenced visits that arrive without a referrer header.

Google Search Console integration

Connect Search Console to a brand to correlate traditional search visibility with AI recommendation trends, and to generate topics directly from the queries your site already ranks for.

AI Share of Voice

A new headline metric measuring a brand's share of visibility across AI answers versus its competitors, with sentence-level brand attribution per AI response.

Fact Checker agent and the Knowledge Base Layer

The Knowledge Base Layer is a brand truth layer that verifies what an AI engine states during a conversation. The Fact Checker agent sits on top of it and flags when an engine says something false about your brand — bringing the agent suite to five: Prompter, Comparer, Introspector, Recommender and Fact Checker.

Brand Perception & Sentiment

How AI talks about a brand, not just whether it does: tone, trend over time, and an AI-perceived SWOT.

Passage-Level Retrieval Analysis

Captures the exact text fragments AI engines cite and maps them back to the source sections they came from, so you can see which passages are being retrieved and which are ignored.

CDN Log Ingestion & Clustering

Process server logs to show which AI crawlers visit a site and which of those visits turn into citations — tying crawl activity to real server traffic instead of estimates.

Shopping Analysis

A new module for tracking product and retailer visibility inside AI shopping conversations. Launched and in active testing with clients.

Enterprise readiness and engine coverage

SOC 2 Type II certification achieved, removing a recurring blocker in enterprise procurement, alongside broader authentication and coverage.

• Enterprise auth: SAML SSO and Microsoft login
• Three new AI engines: DeepSeek, Grok and Microsoft Copilot
• Dedicated in-country proxies for Germany, France, Australia, Czechia and Austria

Content Hub templates and per-topic engines

Reusable templates for generating on-brand articles and briefs, plus the ability to choose which AI engines run on each topic for more granular testing.

Self-serve onboarding in production

Self-serve functionality went live in January, removing the manual setup that every new account previously depended on.

The four-agent suite

Prompter, Comparer, Introspector and Recommender were refined into a coherent suite, each running a strategy adapted to the customer's business type. Personas and multi-turn conversations were sharpened as the core differentiators.

Page Analyser

Explains why a competitor's article gets cited — comparing the content against the topic and scenario it is winning on.

Content Generation

Generate an article or a brief from a topic, a set of keywords and a subject, closing the gap between finding a content gap and filling it.

Competitor head-to-head analysis and query fanout clustering

Direct brand-versus-competitor comparison, and clustering of the fanned-out queries an engine generates from a single prompt.

Citation classification

Every citation is now classified as first-party, competitor, social media, news or community, which required auto-detecting the domains each competitor owns.

Chat with your data

Ask questions of your own visibility data in natural language, including through the Genezio connector for Claude as a skill and over MCP.

New Brand Analysis UI and UX

A rebuilt interface, significantly more advanced than the previous dashboard and designed for deeper analytical exploration.

Fanout queries and statements

Analysis now works at the level of the queries an engine fans out from a prompt, and of the individual statements it makes about a brand.

Three new topic types

Introspector, Recommender and Comparer joined the product as distinct topic types, each probing brand presence from a different angle.

Agency mode

Agency mode in Brand Analysis, for teams running the platform across multiple client brands.

Brand Analysis

The Brand Analysis feature was completed — a periodic report giving customers insight into their brand's visibility across leading LLMs.

Four AI engines

A brand can now be analysed across ChatGPT, Google AI Overview, Perplexity and Google AI Mode.

User management

User management functionality was added to the dashboard.

Three new agents

Industry Bias, Sanitization and Capability Detection joined the evaluation agent library.

Improved existing agents

Fact Checker, Functional Tester, Offtopic, Competitor Advertising and Security were all improved.

Metrics dashboard

A dashboard showing how a chatbot has evolved over time, rather than only how it performs in a single run.

Audit logs

Audit logs were added to the platform.

Evaluation Agents Platform

A new core product, with a dashboard where clients manage knowledge bases, personas, test agent libraries, simulations and detailed reports.

Conversations at scale

Clients can run thousands of conversations in parallel, across multiple languages and configurations.

The first AI Test Agents Library

The first version of our own suite of testing agents, covering three families.

• Personality and behavioural profile agents
• Compliance and industry-specific agents
• Security evaluation agents