Citeloop: tracking whether AI assistants cite your brand
Our own multi client SEO, AEO and GEO platform: a crawler, a findings database and an AI citation scoreboard.
- Client
- Metigital product
- Sector
- SaaS and products
- Year
- 2026
- Built with
- Python
- Postgres
- Supabase
- Next.js
- Vercel
The problem
Search is splitting. Buyers ask an assistant and get one answer with a handful of citations. Ranking reports say nothing about whether your brand is in that answer, and audit tools give a score without telling you which fix matters.
What we built
- Crawler and findings engine covering schema gaps, orphan pages, thin content, duplicate titles and broken links
- Playbooks attached to each finding type: the steps to fix it, the acceptance criteria and the effort involved
- GEO radar that runs a set of buying prompts against multiple AI engines and detects whether the brand is cited
- Client portal with visibility trends, a per prompt scoreboard and a competitor leaderboard
- One database serving every client account with row level security, and quota metering per plan
Where it stands
First production scan covered 442 pages and produced 843 ranked findings, with schema gaps as the single largest fixable group.
Worth knowing
Built so audits stop being an opinion. Every finding is reproducible, and a rescan proves whether a fix actually took.
