LinkStorm

LinkStorm (linkstorm.io) is an AI-powered internal-linking tool for SEOs and publishers. It began as an engagement: François Mommens — a Spain-based founder who already ran two SEO SaaS products, Linkody and IndexCheckr — needed an engineering partner to build his internal-linking product idea. We've been building it since 2023. In 2025 the relationship converted into a formal partnership — the strongest endorsement an engineering client can give.

The problem

Internal linking is high-leverage SEO work that almost nobody does well at scale: it's manual, easy to get wrong, and the first thing that gets dropped. The product had to crawl a site, understand its pages, and place internal links automatically — without the customer touching their CMS.

What we built

• A semantic suggestion engine — every page embedded via OpenAI, stored in per-site Pinecone vector collections, link opportunities scored by semantic similarity. Built in a concentrated sprint in early 2024, long before "just add embeddings" was a cliché.
• Zero-CMS-change injection: a client-side script that places approved links live, marks its own links to avoid re-injection, skips headings and embeds, and ships a diagnostic mode for verifying installation.
• WordPress plugin and Google Search Console integration for performance-weighted link decisions.
• A self-hostable alternative — an open-source Cloudflare Worker (linkstorm-link-injection) for customers who want injection on their own infrastructure. Because lock-in shouldn't be the business model.
• Crawl pipeline with pause/resume, section exclusions, scheduled recrawls, credit metering, and billing webhooks.

Published results from customers, on the product's own stories page: an eCommerce education platform with 14,000+ articles added 43K link suggestions and saw 77% more organic clicks in six weeks; an expense platform reported 10× impressions in three months; a legal platform tripled organic sessions in six months.

How AI runs through it — three different ways

1. In the product: the suggestion engine is an AI system — embeddings plus vector similarity is the only opportunity type LinkStorm generates.
2. In how it's built: 2,100+ commits, with hundreds carrying Claude co-authorship since late 2025. The repo is structured for agent-driven development — the same working method we bring to client work.
3. In how it's operated: churn root-cause analysis runs as an AI investigation across PostHog, Intercom, both databases, analytics and billing — through a custom MCP server we built — with every AI claim cross-checked against the database before anyone acts on it. AI-personalized win-back emails hit a 50% reply rate in their first batch.

Where it stands

LinkStorm runs in production today. We operate it as partners, not vendors — which means when something breaks at 2am, the person who built it is the person who owns it. That's the level of ownership we bring to engagements that aren't ours.