The weekly loop
AI Visibility is a section of the Klype dashboard with three tabs: Scoreboard, Setup and Publish facts. It runs one loop, each week:
- Monitor. Ask the engines your tracked buyer questions and read the answers.
- Plan. Get a short list of next steps based on what the answers showed.
- Produce. Draft posts for the questions where your company was missing, and refresh your GTM plan.
- Publish. Generate files from your saved facts for your own site, so engines have a source to read.
- Measure. Compare the next run with the last one.
Monitor: what the weekly run asks
You track between 5 and 15 buyer questions. Klype can suggest them from your saved category, buyer and competitors, for example “What are the best [category] tools for [your buyer]?” or “What are the best alternatives to [competitor]?”. Suggested questions never name your company, so a question cannot steer the answer. You can edit them and add your own.
Each run sends every question to each engine you have connected a key for:
- ChatGPT, through the OpenAI Responses API with web search
- Perplexity Sonar
- Gemini with Google Search
These calls use your own API keys, which are stored encrypted. Klype does not scrape the consumer apps. A team’s own OpenAI key can count for the ChatGPT engine.
For each answer Klype records the brands named and their order, whether your company is named and at what rank, and the domains cited, including whether your own site was one of them. It keeps a short excerpt of the answer. Weekly runs are off until you turn the schedule on in Setup, and you can also start a run by hand. Manual runs have a weekly cap.
Where answers get facts wrong about your brand
When an answer names your company, Klype compares what it says with the facts saved in your workspace:
- Prices. If an answer states a price for your company that is not one of the prices in your saved pricing facts, it is flagged. This is a fixed rule, and it only runs when you have saved pricing.
- Other claims. The sentences about your company are checked against your saved facts by a model call on one of your connected engines. Differences are flagged with the claim and the saved fact side by side.
A flag means an answer differs from your saved facts. It does not prove the answer is wrong, because the saved fact may be out of date. The next step it creates is to check your facts are current, then publish them.
Plan: next steps from the answers
After a run, the scoreboard lists next steps drawn from what it found. Depending on the result, they include connecting an engine key, tracking at least five questions, correcting what engines say about you, publishing your facts where engines can read them, planning content around the questions you are missing, drafting a post for a missed question, and tracking new names the engines recommend.
Produce: posts and plan for the gaps
A “draft a post” step opens Klype’s Create screen with the missed buyer question as the idea. The GTM plan step opens your GTM plan so your channels answer those questions directly. You review and approve every draft yourself.
Publish: facts for engines to read
The Publish facts tab builds four files from the facts you saved, and nothing else: an llms.txt file, an Organization and FAQ JSON-LD block, and an answers page in Markdown and in HTML. A fact you have not saved becomes a missing line, and the tab lists what is missing so you can add it. Nothing is generated or filled with defaults. You download the files and put them on your own site; Klype does not publish to your site.
Measure: the next run against the last
The scoreboard shows, for the latest run: how many answers named your company, how many cited your site, and your share of voice against the competitors you track. Share of voice here is the share of mentions among your company and the competitors you tracked. Each figure shows its change against the previous run, there is a breakdown by engine, and a week-by-week trend. New brand names that engines recommended and you did not track are listed so you can add them.
What it does not do
- It does not change what any engine answers. It measures answers and prepares your own content and files.
- It does not post or publish for you. Klype never posts as you.
- It only checks answers against facts you have saved. If your saved facts are missing or old, the comparison is too.
- Engine answers vary from one ask to the next, so treat a single run as a reading, not a ranking.
Questions
What does AI say about my brand?
Klype asks ChatGPT, Perplexity and Gemini the buyer questions you track, once a week or when you start a run. For each answer it records which brands were named and in what order, whether your company was named, and which domains were cited, including whether your own site was one of them. It keeps an excerpt of each answer so you can read what was said.
Is there a tool that shows where AI answers get facts wrong about your brand?
Klype's AI Visibility tracker compares what answers say about your company with the facts you saved in your Klype workspace. A price that is not one of your saved prices is flagged by a fixed rule, and other claims about your company are checked against your saved facts by a model call on one of your connected engines. A flag means the answer differs from your saved facts. Check that the saved fact is current before you treat the answer as wrong.
Which AI engines does it ask?
ChatGPT through the OpenAI Responses API with web search, Perplexity Sonar, and Gemini with Google Search. Each runs on your own API key for that provider. Klype does not scrape the consumer apps.
Does Klype change what AI says about me?
No. Klype cannot edit an engine's answer. It measures the answers, lists next steps, helps you draft content, and generates files built from your saved facts for you to publish on your own site. Whether an engine uses them is up to the engine.
Does it publish anything for me?
No. The publish step produces files (llms.txt, Organization and FAQ JSON-LD, and an answers page in Markdown and HTML). You download them and put them on your site yourself.