Skip to main content
# Get Started # Welcome to Orvi AI Orvi AI is a Generative Engine Optimization (GEO) platform that helps marketing teams, e-commerce brands, and agencies understand (and improve) how they appear in AI-generated answers. Your audience isn’t just searching on Google anymore. They’re asking ChatGPT, Claude, Gemini, and Perplexity for product recommendations, category comparisons, and “best for…” lists. These AI conversations shape discovery and buying decisions — and visibility works differently than traditional keyword rankings. Orvi AI shows you when you’re recommended, who you’re compared against, and which sources influence the answer. ## What you’ll find in these docs
  • Get started: Learn what Orvi AI tracks, how GEO differs from SEO, and how to get your first insights fast.
  • Set up your project: How to create a project, craft effective prompts, organize with topics/tags, and choose competitors.
  • Interpret your results: How to read visibility trends, position changes, sentiment shifts, and drill into chats and sources.
  • Take action: How to turn source and competitor insights into content, PR, partnerships, and on-site changes that influence AI answers.
  • Reference: Metric definitions, filtering logic, and platform concepts.
## What Orvi AI tracks (the core GEO metrics) Orvi AI measures AI visibility using three primary metrics:
  • Visibility: How often your brand is mentioned across AI responses for your prompts (think “share of voice” in AI conversations).
  • Position: Where you typically appear when you are mentioned (earlier mentions generally indicate stronger recommendation).
  • Sentiment: How positively (or negatively) the AI describes your brand when it appears.
Under the hood, these metrics are derived from Chats (individual prompt runs) and the extracted entities inside them:
  • Brand mentions: brands detected in the response text (including competitors).
  • Sources: domains and URLs the model relied on (and whether they were explicitly cited).
## Why Sources matter When your visibility or position changes, it’s often because the sources that models rely on changed — new sites started ranking, a publication updated an article, or competitors earned coverage. Sources are one of the most actionable levers in GEO because they influence how models answer (and what they cite). ## How Orvi AI collects data Orvi AI runs your prompts on a recurring schedule and analyzes results over time (AI responses naturally vary day to day).
  • Prompt execution: Orvi AI executes prompts against selected AI models (for example: GPT-5 Chat, Claude Sonnet, Gemini Pro, Perplexity).
  • Structured extraction: Each response is processed to extract brand mentions, positions, sentiment, and sources/citations.
  • Trend analysis: Results are aggregated into time-series charts and tables so you can spot movement over days/weeks, not just one-off answers.
## How Orvi AI is organized (Projects, prompts, competitors) Orvi AI is designed for both in-house teams and agencies:
  • Organization: your company (or agency).
  • Project: a brand/client you’re tracking.
  • Brand prompts (Topics): prompts organized into topics (e.g., “email marketing”, “running shoes”, “protein snacks”).
  • Product prompts: product-focused prompts to understand product recommendation behavior and source patterns.
  • Competitors: competitor domains you choose to track alongside your brand.
  • Tags: optional labels to organize prompts across topics.
## What’s next If you’re new to Orvi AI, start with:
  1. Create a project for the brand you want to track.
  2. Add prompts that mirror real questions customers ask AI (conversational, “best for…” style).
  3. Add competitors so your results have context.
  4. Come back once your prompts have run — then head to Overview and Sources to see your first GEO baseline.
Last modified on January 25, 2026