A teardown of the production pipeline behind every article we ship — research, drafting, the five-dimension quality gate, and the publish-and-monitor loop. Includes the prompt rubric we won't sell.
Read the full teardownMost "SEO briefs" are keyword stuffing in a Google Doc. Here's the rubric we use to brief our drafting agent — it's why our articles read like they were written by a person who actually knows the topic.
Why we abandoned strings-everywhere and put a single TS schema in front of the drafting graph. Includes the migration that took an afternoon and saved a quarter.
We replaced five tools and one freelancer. Two months in, we're up 4× on organic and finally citing on Perplexity.— Sara V. · CEO · LEDGRWORKS Case study
A 12-person fintech that was paying $4.2k/mo across SurferSEO, Jasper, two contractors, and an SEO agency. They kept Webflow.
We watched ChatGPT, Perplexity, and Claude answer the same 700 queries every week for 12 weeks. The patterns in what got cited — and what didn't — are not what the AEO crowd is selling.
Perplexity prefers articles with a specific shape: clear claim → evidence → counter-claim → verdict. Here's the section template we ship now and why it doubled our cite rate.
Your internal links are training signal for AI search. We treat them as a graph problem and rebuild it weekly. Five rules and a Python snippet you can steal.
A clear-eyed taxonomy of seven answer surfaces — Google AI Overviews, ChatGPT, Perplexity, Claude, Gemini, Arc Search, Brave — and which ones moved revenue for our customers last quarter.
A B2B note-taking app with no marketing team. We share the publishing schedule, the topics that hit, and the three pieces that account for 62% of their traffic today.
The biggest release this quarter. We rewrote the project switcher, added IBM Plex Sans Arabic for full RTL articles, and moved the drafting agent to Sonnet 4 — 38% better long-form coherence on internal evals.