elmohq/elmo

What it solves

It addresses the lack of transparent, self-hostable tools for Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO). Most commercial AI visibility platforms are closed SaaS, hiding how metrics are computed and keeping your data on their servers. Elmo gives you a free, auditable way to track how AI assistants mention, cite, and describe your brand across multiple engines, so you can measure your presence, compare competitors, and find actionable opportunities to earn more AI citations.

How it works

You define tracking prompts (questions your customers might ask an AI assistant), either manually or via an onboarding wizard that analyzes your website. A background worker runs each prompt on a schedule against every configured engine, using scraping providers for consumer surfaces like ChatGPT, Google AI Overviews, Gemini, Perplexity, and Copilot, plus direct model APIs like OpenAI, Anthropic, Mistral, and OpenRouter for additional coverage. Each answer is normalized and parsed into text, cited URLs, web searches the engine ran, and model version. A service called Jev reads every answer that mentions your brand or competitors to judge whether a mention really refers to that company, its sentiment (positive, neutral, mixed, negative), and the supporting passage. Everything is stored in PostgreSQL, including raw outputs, so all metrics can be re-derived and audited. Metrics like visibility (share of runs mentioning you), share of voice vs competitors, sentiment scores, position, and citation counts are aggregated and displayed in a dashboard with trends, filters, and reports.

Who it's for

It is built for self-hosting teams that want AI visibility data inside their own infrastructure, agencies offering white-label AEO services with custom branding and SSO, developers and analysts who need auditable metrics via a REST API, TypeScript/Python SDKs, and an MCP server, and anyone priced out of enterprise AEO platforms who still wants credible measurement of how AI engines talk about their brand.

Highlights

• Open source (MIT) and self-hostable via Docker Compose, so you own your data and can inspect every scoring algorithm.
• Tracks visibility, share of voice, sentiment, position, citations, and query fan-out across ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, Copilot, and more.
• Uses an AI judge (Jev) for mention disambiguation and sentiment, with per-answer passages kept as evidence.
• Includes an MCP server and API/SDKs so coding agents like Claude Code and Cursor can query data and manage prompts.
• Offers a shareable reports feature and an optional skill to teach coding agents an evidence-based AEO workflow.