> ## Documentation Index > Fetch the complete documentation index at: https://docs.palagus.ai/llms.txt > Use this file to discover all available pages before exploring further. # Getting Started Set up your project and start understanding how AI platforms present your brand. Palagus analyzes how your brand and competitors appear in AI-generated answers. This guide walks you through the essential setup, from creating your project to interpreting your first results. ## Before you begin A useful Palagus setup answers three questions: * **Which brand do you want to analyze?** * **Who is asking the questions?** * **Which questions matter to your business?** You will define these elements through brands, competitors, personas, topics, and prompts. ## Step 1: Create your project A project contains the brand, competitors, personas, topics, prompts, and AI responses you want to analyze together. **Add brand aliases** Add common variations of your brand name so Palagus can detect them correctly. For example: Palagus, Palagus AI, Palagus AI Visibility, Palagus Tool Only add variations that clearly refer to the same brand. Avoid generic terms that could create false detections. ## Step 2: Add your competitors Add the brands you want to compare with your own. Choose competitors that customers realistically consider when evaluating your product. These may include: * Direct competitors * Established category leaders * Specialized alternatives * Products solving the same problem differently Competitors are used in metrics such as Visibility, Share of Voice, Sentiment, and Position. Start with a focused competitive set. You can add more brands later when they become relevant. ## Step 3: Create your personas Personas represent the people asking the questions you want to monitor. Different audiences may use different language, prioritize different requirements, and receive different recommendations from AI platforms. Examples include: * Marketing manager * Managing director * Procurement manager * Casual Runner * Hobby cook Each prompt in Palagus belongs to one persona. When creating a persona, focus on the person’s role and decision context rather than building an overly detailed fictional profile. A useful persona answers: * What is this person responsible for? * What are they trying to achieve? * Which requirements matter to them? * Which concerns influence their decision? ## Step 4: Create your topics Topics describe the subjects covered by your prompts. Examples: * Integrations * Pricing * Quality A prompt can be connected to multiple topics. Use topics to analyze where your brand performs well and where competitors receive more visibility. ## Step 5: Add your prompts Prompts are the questions Palagus analyzes across AI platforms. A useful prompt should represent a realistic question a person might ask while researching a problem, comparing solutions, or making a decision. ## Step 6: Review your first AI responses Once responses are available, open the individual results before focusing only on aggregated metrics. Review: * Whether your brand is mentioned * Which competitors appear * Where each brand appears in the answer * How the brands are described * Whether the response recommends a particular solution * Which persona and topics belong to the prompt The individual response provides the context behind every Palagus metric. ## Step 7: Understand your results Palagus summarizes the collected responses through four primary metrics. ### Visibility Visibility measures how frequently your brand appears in the analyzed AI responses. A higher Visibility score means that your brand is mentioned in a larger share of the answers. ### Share of Voice Share of Voice measures your brand’s share of the tracked brand mentions. It helps you compare your presence with the competitors included in your project. ### Sentiment Sentiment measures how positively, neutrally, or negatively AI platforms describe your brand. Palagus calculates an aggregated score from the Sentiment of the underlying brand mentions, responses, and prompts. ### Position Position measures where your brand appears within an AI-generated answer: * 10 = At the beginning of the answer * 0  = At the end of the answer If a brand appears multiple times, Palagus averages the Position of its occurrences. The results are then aggregated at response, prompt, and overall level. ## Step 8: Compare personas and topics Avoid evaluating only the overall project score. Use personas and topics to identify where the differences originate. For example, your brand may have: * High Visibility among marketing managers * Low Visibility among technical decision-makers * Positive Sentiment for ease of use * Lower Sentiment for integrations * A strong Position for brand-monitoring prompts * A weaker Share of Voice for reporting prompts These differences are often more actionable than the overall average. ## Common setup mistakes ### Tracking overly broad prompts Broad questions can produce inconsistent results and brands from unrelated categories. Add enough context to reflect the intended customer situation. ### Creating prompts only for your own brand If every prompt explicitly contains your brand name, you cannot reliably understand whether the brand appears in generic discovery and comparison questions. Include non-branded prompts that allow AI platforms to choose which brands to mention. ### Adding too many similar prompts Small wording variations can make your prompt set larger without adding meaningful coverage. Track distinct intents instead of superficial variations. ### Using personas as simple labels Personas should represent different perspectives, requirements, or decision situations. If two personas would ask the same questions for the same reasons, they may not need to be separate. ### Creating too many topics Topics should make analysis easier. Overlapping or overly specific topics create unnecessary complexity. Start simple and expand only when the additional distinction supports a real analysis. ### Focusing only on the headline metrics An aggregated score shows that something is happening. The underlying AI responses help explain why. Always connect changes in metrics back to specific prompts and responses. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.