The problem
Great content is slow and expensive to produce at volume
Writers spend hours on research, outlining, and SERP analysis before writing a word. Output is inconsistent in voice, thin on internal links, and rarely mapped to search intent. Scaling means hiring — or shipping mediocre content that never ranks.
The Content Engine compresses the research-to-draft cycle from days to minutes while keeping a human as the editor, not the typist.
Primary objective
Produce ranking-ready, on-brand drafts at 10× the pace — without sacrificing quality or accuracy.
How it works
Brief → research → outline → draft → QA
1 · Brief intake
Keyword, intent, audience, brand voice profile
→
2 · SERP research
Scrape top results, extract headings, entities & gaps
→
3 · Outline
Intent-mapped H2/H3 with entity coverage targets
4 · Draft
On-voice long-form with internal links inserted
→
5 · QA gate
Fact-check, plagiarism, readability, brand-safety
→
6 · Human review
Editor approves & publishes to CMS via API
What it does
- ✓ SERP-driven outlines
- ✓ Brand-voice fine-tuning
- ✓ Automatic internal linking
- ✓ Entity & topic coverage scoring
- ✓ Meta title & description generation
- ✓ Schema.org markup suggestions
Built with
- ✓ Claude for drafting & editing
- ✓ RAG over your site & style guide
- ✓ SERP + rank-tracking APIs
- ✓ CMS connectors (WordPress, Webflow, Contentful)
- ✓ Human-in-the-loop approval queue
Guardrails
- ✓ Citation-backed claims only
- ✓ Plagiarism & AI-detection checks
- ✓ No publish without human sign-off
- ✓ Brand-safety & tone filters
- ✓ Full revision audit trail
Under the hood
SERP-grounded outline generation
Research the live SERP, then draft an outline the model must ground in real competitor headings and entity gaps — not its imagination.
engine/outline.py
PYTHON
from anthropic import Anthropic from serp import fetch_top_results, extract_entities client = Anthropic() MODEL = "claude-sonnet-4-20250514" def build_outline(keyword: str, brand_voice: str) -> dict: # 1. Ground the model in the live SERP, not guesses top = fetch_top_results(keyword, n=10) headings = [h for r in top for h in r.headings] entities = extract_entities(top) # entities competitors cover prompt = f"""You are an SEO content strategist. Keyword: {keyword} Competitor headings: {headings} Entities to cover: {entities} Write a JSON outline (H2/H3) that covers every entity above, fills gaps competitors miss, and matches this brand voice: {brand_voice}. Reply with ONLY JSON.""" msg = client.messages.create( model=MODEL, max_tokens=1500, temperature=0.3, messages=[{"role": "user", "content": prompt}], ) return parse_outline(msg.content[0].text) # validated downstream
Outcomes
10×
Faster brief-to-draft
~70%
Lower cost per article
100%
On-brand, human-approved
1
Editor runs a whole team's output
Want this running on your content pipeline?
Deploy it standalone, or bundle it with the rest of the SEO & Marketing suite.
More SEO & Marketing solutions