Make website content discoverable again — semantic, citation-ready site search

Many mid-market and SMB publishers, product documentation teams, and content-heavy sites struggle to find and repurpose legacy content because built-in site search remains largely keyword-based, returns poor relevance, and offers no easy way to cite or surface source material for updates; this problem affects roughly 2,000,000 candidate sites. The practical outcome is wasted editorial hours, missed SEO value from evergreen articles, and slower product onboarding for companies that rely on internal knowledge, with site owners often unable to quantify the loss. You could build a semantic, citation-ready site search platform that combines embeddings and RAG for relevance, automated provenance (clickable citations and source snippets), incremental indexing for freshness, and first-class plugins for WordPress, Shopify, and Ghost plus a developer API and analytics dashboard; target ACV of about $4,000 per site aligns with an $8.0B addressable market (2M x $4,000) and the business has been scored 88/100 for market attractiveness and 90/100 for revenue potential. The technical stack would include a vector database, managed LLM inference with cost controls, and content workflows for repurposing suggestions—real engineering work but feasible with current toolchains. This market is attractive now because AI semantic search adoption (embeddings + RAG) materially increases internal search relevance and publishers are prioritizing content repurposing, while CMS plugin ecosystems make distribution far easier than building bespoke integrations. To stand out you should emphasize provable citation accuracy, low total cost of ownership (predictable inference billing), seamless CMS integration, and clear ROI metrics for editors and SEO, while realistically planning for medium competition, upfront customer acquisition costs, ongoing model and hosting expenses, and the need for strict relevance tuning and privacy controls.