Most AI writers start with a keyword and hallucinate from there. We built a research intelligence pipeline that reads your sources, builds a knowledge base from your niche, and generates articles grounded in verifiable information.
Every source you add contributes to a private knowledge base. Blog posts, YouTube transcripts, Reddit discussions, podcast episodes. When we generate an article, we don't start from zero. We query your knowledge base to find relevant context, then cross-reference it with the competitive landscape.
We understand meaning, not just keywords. Ask about "monitoring server performance" and we'll find articles about observability, APM tools, and infrastructure dashboards.
For each article, we run 5-7 targeted queries derived from your brief's keywords, questions people ask, and related entities. Not one search. A research strategy.
We index your own site too. So when we write about a topic you've covered before, we can reference your existing work and suggest internal links naturally.
Every fact in every article traces back to a source document. You can verify claims, check context, and see exactly where information came from.
We read your published content and measure how you actually write. Not a vague "tone" setting. Quantified patterns that distinguish your writing from everyone else's.
Length variation and burstiness
How you open and structure sections
Common, domain, and rare word balance
First person, second person, passive usage
Your repeated phrases and metaphors
How you move between ideas
Concrete examples vs. abstract claims
Facts per sentence, how tightly you write
The result is a deterministic style guide, roughly 17,000 words, generated entirely from statistical analysis of your corpus. No AI interpretation, no hallucinated style descriptions. Pure measurement.
When a brief mentions "MCP servers", we don't just treat it as a keyword. We resolve it to a structured entity, discover related concepts (protocol handlers, tool interfaces, stdio communication), and use that understanding to find better research and generate more informed content.
Every brief identifies 5-8 key entities: technologies, organizations, people, concepts. Each one gets resolved to a knowledge graph identifier.
From those entities, we discover sibling concepts and parent categories. An article about "Newton meters" automatically gets context about torque, force measurement, and related units.
Every published article includes a provenance chain: which facts came from which sources, which entities were referenced, and a full research audit trail. This is machine-readable and embedded as structured data for search engines.
Four independent signal generators read your sources every day: trending topics, decaying coverage, untapped angles, and audience gaps. Diversity selection stops the same story filling your inbox five times.
Source monitoring →Every brief is validated against keyword volume, difficulty, monthly trend, "people also ask" questions, and your own Search Console data. We do not pitch topics nobody searches for.
Content briefs →Each topic is scored against the existing landscape: how many guides already rank, how good they are, what video coverage exists, and where the gaps are. Originality scoring gates how we pitch the angle.
Ingestion, content reading, synthesis, enrichment, delivery, and article production each run as their own stage with their own model choice. The right tool for each job, not one prompt doing everything badly.
See the pipeline →A REST API and an MCP server expose every tool: query your corpus, list briefs, generate articles, manage sources. Drop it into Claude Desktop, Claude Code, or Cursor with one config block and search your private knowledge base from your editor.
API & MCP docs →Every claim traces back to a real source. No made-up statistics, no phantom citations. You can check every reference before publishing.
Articles match your measured writing patterns. Sentence cadence, vocabulary, paragraph structure. Your readers won't notice the difference.
Structured data embedded in every article. Entity annotations, source citations, and provenance metadata that search engines can read and trust.
Free tier monitors up to 5 sources with weekly briefs. Paid plans add articles, voice matching, and WordPress publishing.