How Content Generation Works

Wyrote uses a two-layer pipeline to go from zero to published content. First, a strategy pipeline creates your content plan. Then, a content generation pipeline writes each individual article. This guide explains both processes so you understand exactly what happens behind the scenes.

The Strategy Pipeline (5 Phases)

When you generate a strategy for an offering, Wyrote runs through 5 phases to build a complete content plan:

Phase 1: Keyword Discovery

An AI agent generates 100+ relevant keywords for your offering. It considers:

• Your company profile and industry
• The offering description
• Target countries and language
• Common search patterns in your sector

Token cost: 8 tokens

Phase 2: Market Analysis

Real SEO metrics are fetched for every keyword from DataForSEO:

• Search Volume — Monthly search frequency
• Keyword Difficulty (KD) — How hard it is to rank (0–100)
• CPC — Cost per click in paid ads
• Competition Level — How competitive the keyword is
• Search Intent — Informational, commercial, or transactional

Token cost: 8 tokens per 50 keywords

Phase 3: SERP Analysis

The system performs competitive gap analysis by examining:

• Top-ranking pages for your keywords
• People Also Ask (PAA) questions
• Related searches
• Content patterns of competitors

Phase 4: Topic Clustering

Keywords are grouped into a pillar-cluster model:

• Pillar topics — Broad, high-level topics that cover a wide area
• Cluster topics — Specific sub-topics that link back to the pillar

This creates a content architecture that builds topical authority in search engines.

Token cost: 12 tokens

Phase 5: Content Planning

An AI planner creates article ideas for each cluster:

• Article titles
• Content types (pillar article, support article, comparison, guide, etc.)
• Target keywords per article
• Publishing schedule spread across the month

Token cost: 15 tokens

The Content Generation Pipeline (8 Stages)

When you approve an article (or it's auto-approved), the generation pipeline writes the full article through 8 sequential stages:

Stage 1: Research

• Scrapes top-ranking competitor pages for the target keyword
• Collects SERP data, PAA questions, and related searches
• Extracts authoritative sources from real data (government, academic, industry sites)
• Gathers information to ensure the content is data-backed

Stage 2: Outline

An AI agent creates a detailed article outline:

• Section headings and structure
• Key points to cover per section
• Logical flow and information hierarchy
• Target word count per section

Stage 3: Section Writing

Each section is written individually, in parallel (up to 5 concurrent sections):

• Text-only content (no images at this stage)
• Uses research data and outline as guides
• Incorporates verified authoritative sources
• Links 1–2 external sources per section where appropriate

Stage 4: Enrichment

The enrichment stage adds structural elements:

• Internal links — Links to your own pages (from your sitemap)
• CTA elements — Calls to action linking to your offerings
• Validates link density and distribution

Stage 5: AI Pattern Detection

A specialized AI model scans the content for detectable AI writing patterns:

• Identifies repetitive structures, overused transitions, and formulaic phrasing
• Flags sections that read as obviously AI-generated
• Creates a rewrite plan for flagged sections

Stage 6: Targeted Rewrite

Based on the detection results, a second AI model rewrites only the flagged sections:

• Varies sentence structure and length
• Adds natural language patterns
• Maintains factual accuracy while improving readability
• Preserves the content's SEO structure

Stage 7: Fact-Checking

An AI fact-checker reviews the article:

• Verifies claims against research data
• Checks that statistics and data points are accurate
• Ensures source attributions are correct
• Flags any unsupported claims

Stage 8: Editing

A final editing pass polishes the content:

• Grammar and style corrections
• Tone consistency with your voice profile
• Applies your workspace content rules and offering-specific rules
• Ensures SEO elements (headings, keyword usage, meta data) are optimized

Post-Pipeline: Image Generation

After the article is marked as ready:

• Section images are generated for sections marked for visual content (infographics, section images)
• Featured/hero image is generated as a fire-and-forget process
• Images are created based on the final, verified content to avoid inaccurate visuals

Watching Generation Live

When you approve an article, you're redirected to the content view where you can watch the pipeline in real-time:

• A step tracker shows which stage is currently running
• Section counter displays progress (e.g., "Section 3 of 8")
• Content appears progressively as sections are completed
• If a stage fails, the error is displayed with the failed stage highlighted

Token Costs Summary

Next Steps

• Daily Content Workflow — Approve content and manage your daily queue
• Content View — Understanding the article page and generation progress
• Voice Profile — Configure your writing voice for more natural content