Answer engines retrieve passages, not pages. Trilar splits your content the way they do, scores every section and shows exactly what to add or rewrite.
Trilar splits the main content of every page at its headings and scores each section from 0 to 100: length between 40 and 120 words, a first sentence that names the topic and stands on its own, concrete numbers and facts, and a direct answer under question headings.
AI search splits a question into sub-queries and retrieves a passage for each. Trilar has an AI model write those sub-queries for a page and marks which passage answers each one fully, partly or not at all. The gaps are the content to add.
How much of a page is its own content and how much is repeated template text. Thin pages with little own text are hard to cite and easy to ignore.
The trust signals of the whole site, without AI: imprint, contact, about and privacy pages and how widely they are linked, your organisation in the schema, and per page author, date, author profiles and editorial review. On request, an AI model rates single pages for experience, expertise, authoritativeness and trust with evidence and fixes.
For topics your site lacks, Trilar writes a content brief from your Search Console queries and the pages that rank. For images without alt text it suggests one per image, page by page.
SEO makes sure search engines can crawl, understand and rank a site. GEO (Generative Engine Optimization) is about being cited as a source in AI-generated answers. AEO (Answer Engine Optimization) shapes content so that single sections answer a question directly and on their own, which is what answer engines quote. Trilar checks all three in the same crawl.
Answer engines quote sections, not whole pages. Trilar splits every page at its headings and scores each section: length, whether the first sentence names the topic and stands on its own, concrete facts and numbers, and direct answers to question headings.
Without AI, Trilar checks the trust signals of the whole site: imprint, contact, about and privacy pages, the organisation in your schema, and per page author, date, author profiles and editorial review. On request, an AI model rates single pages for experience, expertise, authoritativeness and trust, with evidence from the text and concrete fixes.
Trilar uses Claude by Anthropic through its commercial API. AI features only run when you start them, and results are stored with the crawl so you do not pay twice. Anthropic does not use API inputs to train its models by default. Everything else, from the crawl to the action plan, runs without AI.
Trilar is in early access. Plans are based on the number of crawled URLs per month plus credits for AI features. Write to us and we will set up access and discuss the right plan.
Trilar crawls every URL like Google, renders JavaScript, connects Search Console and Analytics, and turns thousands of findings into a short list of fixes, most expensive first.
geoAI search engines answer with a few sources. Trilar shows whether yours is among them, which domains win instead, which of your pages should have been cited and what keeps it from being quoted.
Early access includes onboarding: we set up your account and go through the first crawl with you.