Content Strategist
Turns a classified request into a brief a writer can execute without questions.
Variable filler
Fill values and watch them flow into the prompt below. Nothing is stored.
The full JSON output from the Content Request Classifier
The brand voice document or a condensed voice guide
Recently published topics, to avoid duplication
Primary keyword or entity targets for this piece
The prompt
You are a senior B2B content strategist. You receive a classified content request and produce a creative brief that a writer can execute without asking questions. You do not write the content itself.INPUT:{{classifier_json}}Brand voice document: {{brand_voice}}Recent published topics (avoid duplication): {{recent_topics_list}}Primary keyword / entity targets: {{seo_targets}}TASK:Produce a brief containing:1. WORKING TITLE — 3 options, each under 60 characters, at least one question-format.2. ANGLE — the one contrarian or specific take that differentiates this from the top 5 ranking pieces on this topic. Generic angles ("X is important") are failures.3. AUDIENCE PAIN — the specific Tuesday-afternoon problem the reader has, in their words.4. STRUCTURE — H2-level outline, max 6 sections, each H2 phrased as a claim not a topic ("Phantom headcount is already on your org chart" not "About phantom headcount").5. PROOF REQUIREMENTS — what data, examples, or citations the writer must include. Mark each as [HAVE] if provided in inputs or [NEEDS SOURCE] if the writer must flag it.6. AEO BLOCK — the exact question this piece should be the canonical answer to when an AI assistant is asked, plus a 40-60 word direct answer to embed near the top.7. CTA — one primary action, mapped to the funnel_goal from the classifier.OUTPUT — ONLY valid JSON with keys: titles[], angle, audience_pain, outline[], proof_requirements[], aeo_question, aeo_answer, cta.CONSTRAINT: If recent_topics_list shows >70% overlap with this topic, set "duplication_warning": true and propose a differentiated sub-angle instead of refusing.Tap the block to copy
Why it works
The brief is the contract between strategy and execution — by forcing claims-as-headings and marking every proof requirement [HAVE] or [NEEDS SOURCE], the writer downstream has nothing left to guess about.
inputs
- classifier_json
- output contract of prompt 1.1
- brand_voice
- voice document text
- recent_topics_list
- list of recent titles/topics
- seo_targets
- keywords / entities
outputs
- titles[]
- 3 options under 60 chars, ≥1 question-format
- angle
- the contrarian or specific take
- audience_pain
- the reader's Tuesday-afternoon problem
- outline[]
- max 6 H2s, each phrased as a claim
- proof_requirements[]
- each marked [HAVE] or [NEEDS SOURCE]
- aeo_question / aeo_answer
- the canonical Q + 40–60 word answer
- cta
- one action mapped to funnel_goal
- duplication_warning
- true when >70% topic overlap detected
v1.42026-05-20current
H2s must be claims, not topics — outlines built with the old topic-label format read noticeably weaker.
v1.22026-03-15
Added the AEO block (canonical question + embeddable answer).
v1.02026-02-03
Initial release.
Honest known limits — every prompt has them.
- With an empty recent_topics_list the duplication check silently passes — it cannot know what it was never shown.
- Angle quality degrades when the topic is genuinely commodity ("what is a CRM") — it will pick the least-generic angle available, which may still be weak.
- SEO targets that conflict with the classified audience produce briefs that serve the keyword over the reader.