CitedAI
Your site is invisible in AI search and your SEO agency has no answer. CitedAI audits your brand's citation rate across ChatGPT, Perplexity, and Google AI Overviews, then tells you exactly which pages to rewrite and how.
The Idea
Gartner predicts a 25% decline in traditional search engine volume by 2026. Perplexity's query volume grew 239% in nine months. Google AI Overviews now appear on 18% of commercial queries, up from 8% a year ago. The traffic is shifting, and most businesses are completely unprepared. CitedAI is an AI search visibility tool: it runs your target keywords through ChatGPT, Perplexity, and Google AI Overviews, tracks which brands get cited in the answers, and gives you a Citation Score. Then it analyses your existing pages to tell you why you're not being cited and what to change. The output is a prioritised rewrite plan, section by section, based on the specific structural and factual signals that AI engines use when selecting sources.
Why Now
The traffic shift is measurable and accelerating right now. Perplexity handled 780 million queries per month by May 2025, nearly triple its August 2024 volume. Google AI Overviews are triggering on commercial intent queries at more than double the rate of a year ago. Adobe Digital Insights has documented consistent organic click decline as zero-click AI answers replace traditional search results. The problem: the entire SEO industry built its skillset around Google's ranking algorithm. None of those skills map directly to AI citation. Businesses are watching their organic traffic charts tilt south and their agencies are shrugging. The GEO (Generative Engine Optimization) discipline is real and documented by Search Engine Land, Similarweb, and Gartner, but the tooling for SMEs does not exist yet. Enterprise GEO services start at £5,000 per month. The self-serve gap is wide open.
How to Build
The core product is a citation tracking and content gap tool. The user enters their brand name, domain, and 10 to 20 target keywords. CitedAI queries ChatGPT (via the API), Perplexity (via their API), and scrapes Google AI Overviews for each keyword using a headless browser. It records which sources are cited in each answer and whether your brand or domain appears. This produces a Citation Score: the percentage of relevant queries where your brand is cited across all three platforms. For queries where competitors are cited but you are not, CitedAI fetches the cited competitor pages and compares them structurally against your corresponding pages. The documented citation patterns: Q and A formatted content, clear author credentials, specific statistics with named sources, structured data markup, and concise direct answers to question-based queries. The tool outputs a page-by-page rewrite checklist. Optional: paste your page content in and CitedAI rewrites underperforming sections using the citation-friendly format. Stack: Next.js, Vercel, Stripe, OpenAI API, Perplexity API, Playwright for Google scraping.
Revenue Model
£49 per month: track up to 25 queries across three AI platforms, monthly Citation Score report, basic gap analysis. £149 per month: unlimited query tracking, competitor citation comparison, automated page rewrite suggestions, weekly monitoring alerts. Agency tier at £349 per month: multi-client management, white-label PDF reports, bulk keyword monitoring. The pricing anchors directly against the alternatives: enterprise GEO consultants at £5,000 per month, or doing nothing and watching organic traffic continue its decline. For any business generating meaningful revenue from organic search, £149 per month to understand and fix AI search visibility is a straightforward spend. Target 150 active subscribers at the Pro tier within 6 months: £22,350 MRR. Distribution is unusually direct: every digital marketing agency in the UK has clients asking "why is our traffic down" and no credible answer. Sell to agencies first and let them bring their client book.
Effort
1 to 2 weeks to a usable v1. Day 1 to 2: the query runner, sending target keywords to ChatGPT and Perplexity APIs and parsing responses for brand and domain citations. Day 3: Google AI Overviews scraping via Playwright. Day 4: Citation Score calculation and the structural comparison logic against cited competitor pages. Day 5: the content gap output, formatted as a checklist per page. Day 6: Stripe billing and basic user onboarding. Day 7: the content rewriter module using Claude or GPT-4. The citation parser is the interesting engineering problem: LLM responses are unstructured, so you need a reliable extraction layer for source attribution. Build that robustly before adding features. A consistent, accurate Citation Score is the whole product.
Risk
Two real risks. First, measurement instability: AI models are updated constantly and citation patterns change with each model version. A Citation Score accurate in February may not reflect the same behaviour in May. Frame it as a directional signal, not a precise metric, and be transparent about this. Second, unsettled best practices: GEO recommendations are still being worked out by the industry. What increases citation rates today may not work in 12 months. This is actually a retention advantage (customers stay subscribed to track the moving target) but it means your rewrite recommendations must be grounded in documented structural signals, not speculation. Source your methodology from published research and update it publicly when the evidence shifts.
The traffic shift is happening right now and the tooling to respond to it barely exists. Every business with meaningful organic search revenue has a direct problem that CitedAI solves. The pricing is defensible, the build is scoped for two weeks, and distribution through SEO agencies is unusually efficient: one agency conversation unlocks 10 to 20 client accounts. The measurement instability risk is real but manageable, and the moving target actually creates retention. Build this.