> ## Documentation Index > Fetch the complete documentation index at: https://docs.trymudra.com/llms.txt > Use this file to discover all available pages before exploring further. # Content Lab > Morphiq Labs's workspace for turning tracked prompts into citation-ready pages that AI models can easily quote. Content Lab is Morphiq Labs's workspace for turning **tracked prompts** into **citation-ready pages**. It takes the sources AI models already cite, adds fresh research, and ships a structured article that's easy for models to quote. Content Lab generation steps *** ## What it is * **Prompt-aligned:** Every page starts from a tracked prompt (or cluster) you care about. * **Source-aware:** Uses the same domains/URLs AI models cite, not random web pages. * **Structured for models:** One clear H1, logical H2/H3s, short paragraphs, optional schema suggestions. When you open Content Lab for a prompt, Morphiq Labs: 1. Runs that prompt across supported AI models. 2. Captures their answers and cited sources (domains + URLs). 3. Uses those sources plus fresh research to build a better, more complete answer on your domain. No diffing against your existing pages — the goal is a **stronger, clearer version** of what models already try to answer. *** ## How it works (behind the scenes) 1. **Ingest sources** * Pulls the citations tied to the tracked prompt. * Filters and deduplicates; keeps up to 10; needs at least 2 to proceed. 2. **Scrape sources (Firecrawl)** * Grabs the main content from each URL (markdown + links). * Keeps going even if some URLs fail; stops only if fewer than 2 succeed. 3. **Gap analysis** * Compares scraped content against the tracked prompt. * Flags missing angles, data gaps, weak formats, and suggests search queries. 4. **Live research** * Runs focused searches (recent data, original studies, expert quotes). * Adds a small set of additional sources with key insights. 5. **Draft generation** * Produces a 1,200–1,600 word article with: * One H1 and direct-answer H2s * TL;DR right under the title * Short paragraphs (2–4 sentences, 50–75 words) * Comparison table when the prompt is comparative * Bottom Line + FAQ (3–5 Q\&As) * Includes author name/title from the brand profile and tracks word count, sections, and sources. Result: A complete, GEO-optimized article saved to your campaigns with all sources (scraped + research) logged. *** ## What you see in the UI * **Step 1:** Pick content type — General Blog Post, Listicle, How-To Guide, Comprehensive Guide, or Comparison. * **Step 2:** Pick the tracked prompt (by category or list). * **Step 3:** Pick the target audience/ICP. * **Step 4:** Review & toggle citation sources (needs at least 2). * **Step 5:** Watch generation progress (validate → scrape → analyze gaps → research → draft). * Auto-redirects to the campaign page when ready. *** ## What it produces * **Core article / pillar page** — Direct answer to the tracked prompt with clear sections. * **Supporting Q\&As** — FAQ blocks aligned to how users and models split the topic. * **Research-enriched sections** — Uses primary studies, reports, and fresh data instead of generic claims. * **Schema-ready structure** — Headings, FAQs, tables, and lists that are predictable for models. * **Saved campaign** — Article, metadata (title, sections, word count, author), and all sources tracked. *** ## How the pipeline works Behind the scenes, Content Lab runs three stages to produce each draft: * **Content Quality:** Enforces clear titles, upfront answers, evidence-backed claims, and E-E-A-T signals. * **Content Structure:** Keeps the heading hierarchy clean, paragraphs short, and suggests tables/lists/FAQs. * **Citation & Research:** Spots claims that need proof, pulls recent high-authority sources, and keeps citations tidy. You keep final voice and accuracy; the pipeline handles the heavy lifting so the draft is safe to cite and easy to reuse. *** ## What to prepare before you start * The tracked prompt (or cluster) you want to win. * At least **2 citation sources** tied to that prompt (auto-loaded if available). * Brand profile basics: brand name, description, target ICP, unique value prop, author name/title. * Optional: Any non-public data or examples you want woven in (you can paste them into the prompt context). *** ## Why it matters * You publish pages that **match how models already answer** — with better depth, fresher data, and clearer structure. * Every section is designed to be **quotable and scannable** by AI models. * Sources are tracked, research is fresh, and output is ready for GEO-focused distribution.