> ## Documentation Index > Fetch the complete documentation index at: https://docs.oppin.ai/llms.txt > Use this file to discover all available pages before exploring further. # Perception > Hear how AI describes you. Prompts to paste into Claude or any assistant connected to Oppin. Paste a prompt into an assistant connected to Oppin ([how to connect](https://docs.oppin.ai/mcp/overview.md)) and replace the parts in \[brackets] with your own. In Oppin, the **MCP** tab of the **Data & MCP** page has the same prompts with your brand and a competitor already filled in. ## Quick questions One question, one answer. Paste it as it is. ### How does AI describe us? *Quick question* The distinctive words AI uses about you, each with an example. ```text wrap theme={null} Using Oppin, how does AI describe us? ``` ### Show me the most negative answer about us. *Quick question* The answer, the engine, the question and the sources behind it. ```text wrap theme={null} Using Oppin, show me the most negative answer about us. ``` ## Workflows A few steps and one useful result. ### What's dragging sentiment down *Workflow · Interactive page (quote wall)* Find the specific answers and pages behind a sentiment drop. ```text wrap theme={null} Using Oppin, why did our sentiment drop? Build a quote wall: the most negative answers about us, with the negative phrase highlighted, the assistant and question each came from, and the page behind it. Then tell me which pages keep showing up. ``` * **You get:** A wall of real quotes where the problem is impossible to miss, when the drop started, the recurring themes and the pages behind them. * **Why it helps:** An average hides the one page dragging hundreds of answers down. ### Positioning check *Workflow · Ranked list (claim by claim)* See whether AI repeats your messaging, contradicts it, or ignores it. ```text wrap theme={null} Using Oppin, here's our positioning: [paste]. For each claim, tell me whether AI reflects it, contradicts it, or doesn't mention it. Quote the answers. ``` * **You get:** A claim-by-claim table (reflected / contradicted / absent) with quotes, and the top three gaps. * **Why it helps:** Shows exactly where messaging isn't reaching AI. ### Side-by-side answers *Workflow · Interactive page (one column per assistant) · For leadership* Feel what a buyer sees when they ask AI about your category. ```text wrap theme={null} Using Oppin, pick one of our most important questions and show me how ChatGPT, Gemini and Perplexity (or whichever assistants we track) each answered it, side by side. Highlight our brand in one color and competitors in another, and list the sources under each answer. ``` * **You get:** Real answers next to each other: who each assistant recommends first, how you're described, which sites each one leaned on. * **Why it helps:** A number says you're at 30% visibility; this shows the answer you're missing from. The fastest way to make leadership care. ### Answer explorer *Workflow · Interactive page (filterable list)* Give anyone on the team a way to browse what AI actually says, without opening the dashboard. ```text wrap theme={null} Using Oppin, build an interactive page of our last 50 AI answers: filters for assistant, question, tone and whether we're mentioned, a preview of each with our brand highlighted, and the full text with sources for the 10 most negative ones. ``` * **You get:** A snapshot someone can click through in a meeting: what AI says about you, in its own words. * **Why it helps:** "See what buyers see" for people who will never log into a dashboard. The previews come in one call; full answers are the expensive part, so it opens only a few. ### Track whether AI gets corrected *Workflow · Ranked list (wrong claims) + questions added on approval* Turn each wrong claim about you into a question that shows when AI gets it right. ```text wrap theme={null} Using Oppin, find answers where AI says something wrong about us (price, features, availability). The correct details are: [paste or link]. For each wrong claim, suggest a question that tests it directly ("Does [brand] have a free plan?"), tagged "fact check", and add the ones I approve so we can see when the answer changes after we fix the source. ``` * **You get:** Each wrong claim with the answer it came from and the page behind it, plus one tagged question per claim that shows when the answer is fixed. * **Why it helps:** A wrong answer is the most upsetting thing Oppin shows. This lets you watch it get fixed. * **Changes tracking:** needs a connection with write access. The assistant shows the changes and waits for your yes. * **Take it further:** see [Fix what AI gets wrong](https://docs.oppin.ai/mcp/use-cases/citations.md#fix-what-ai-gets-wrong) and [Positioning check](https://docs.oppin.ai/mcp/use-cases/perception.md#positioning-check). ### News watch *Workflow · Answer (the question list) + questions added on approval, then a two-week check* Track how AI talks about a news story while it matters, then stop. ```text wrap theme={null} Using Oppin, there's a news story about [issue]: [link]. Suggest questions that test how AI talks about it (is [brand] safe, what happened with [brand] and [issue], alternatives to [brand]) on every assistant we use, tagged "watch: [issue]", and add them after I approve. In two weeks, show me how the answers changed and stop the ones we no longer need. ``` * **You get:** A tagged set of questions on the story, and two weeks later how the answers moved, with the ones to stop. * **Why it helps:** A reputation moment is when leadership opens the tool. Tracking starts and ends with it: stopping keeps the history and gives the plan its room back. * **Changes tracking:** needs a connection with write access. The assistant shows the changes and waits for your yes. * **Note:** Seasons work the same way: "Using Oppin, add a Holiday gifting topic with 10 gift-guide questions in \[US, UK, Canada], tagged holiday. On \[Jan 5], stop them." The later step is a second request, or a scheduled task in your assistant. * **Take it further:** see [What's dragging sentiment down](https://docs.oppin.ai/mcp/use-cases/perception.md#whats-dragging-sentiment-down) and [Campaign or launch impact](https://docs.oppin.ai/mcp/use-cases/pulse.md#campaign-or-launch-impact). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.