Writing

Technical Blog Post Outline

Turns a topic and audience into a structured outline with per-section word budgets, a stated thesis, and an explicit list of what the post will not cover.

Prompt template

You are a technical editor at an engineering publication. Your job is to
produce an outline, not prose. You will be judged on structure and scope
discipline, not on eloquence.

## Inputs
Topic: {{TOPIC}}
Audience: {{AUDIENCE}}
Target length: {{WORD_COUNT}} words
Primary takeaway the reader should leave with: {{KEY_TAKEAWAY}}

## Task
1. Write a one-sentence thesis. It must be falsifiable — a claim someone
   competent could disagree with, not a description of the topic.
2. List 3 things a reader of the stated audience already knows. The outline
   must not spend words re-explaining them.
3. Produce 6-9 H2 sections. For each: the heading, a one-line purpose, the
   specific evidence or example it needs, and a word budget.
4. Mark exactly one section as the load-bearing section — the one that, if
   cut, would make the thesis unsupported.
5. List 3 topics adjacent to this one that the post will explicitly NOT cover,
   with a one-line reason for each.

## Constraints
- Word budgets must sum to within 10% of {{WORD_COUNT}}.
- Do not propose a "Conclusion" section that only restates earlier points.
- Do not use the words "delve", "landscape", "journey", or "unlock".
- Do not write any body copy. Headings and notes only.
- If {{TOPIC}} is too broad to support one thesis at {{WORD_COUNT}} words,
  say so in a single line and propose a narrower topic instead of outlining.

## Output format
Markdown, in this exact order:
**Thesis:** ...
**Assumed knowledge:** bulleted list
**Outline:** numbered list of sections, each with `Purpose:`, `Evidence:`,
`Words:` sub-bullets, and `[LOAD-BEARING]` on exactly one heading
**Out of scope:** bulleted list

Why this works

Three techniques do the work here. Forcing a falsifiable thesis blocks the model's default behavior of producing a topic survey with no argument. Per-section word budgets that must sum to a target act as a numeric constraint the model can self-check against, which reliably prevents fifteen-section sprawl. The explicit out-of-scope list is negative space specification — naming what to exclude suppresses tangents far more effectively than asking for focus.

Example output

Example

**Thesis:** Vector search is usually the wrong first fix for a bad RAG
pipeline; retrieval quality is dominated by chunking decisions made upstream.

**Assumed knowledge:**
- What an embedding is
- Basic RAG architecture (retrieve, then generate)
- That cosine similarity ranks results

**Outline:**
1. The symptom teams misdiagnose — Purpose: establish the failure pattern.
   Evidence: a query returning topically correct but useless chunks.
   Words: 250
2. Why chunk boundaries decide recall [LOAD-BEARING] — Purpose: prove the
   thesis. Evidence: same corpus, two chunking strategies, different results.
   Words: 500

**Out of scope:**
- Embedding model selection — a smaller effect than chunking at this stage
- Reranking — worth a post of its own
- Vector database benchmarks — vendor-specific and quickly outdated

Usage tips

Run this before you write, and treat the out-of-scope list as binding. Most drafts bloat because a tangent felt relevant mid-paragraph; having pre-committed to excluding it makes the decision once instead of five times.

If the model returns a thesis that nobody would argue with (“RAG systems have several components”), reject it and re-run with the instruction that the thesis must be something a competent engineer could dispute. That single correction improves the whole outline.

Variation

For a tutorial rather than an argument piece, replace the thesis instruction with: “Write a one-sentence statement of what the reader will be able to do at the end that they could not do at the start.” Everything else works unchanged.

Browse library

Bullet Notes to Voice-Matched Draft

Converts rough notes into a first draft that imitates a supplied writing sample, with strict rules against inventing facts the notes do not contain.

Self-Critique Line Edit Pass

A two-phase editing prompt: the model first diagnoses concrete problems in your draft against named criteria, then rewrites only the passages it flagged.

The AI signal without the hype — new models, tools worth your time and what actually shipped. One email, no vendor pitches.

Choosing an AI stack for your team?

We help companies pick the right AI tools, wire them into existing systems and avoid the ones that quietly do not scale. Tell us what you are trying to build and we will tell you what we would use — no charge for the conversation.