ChatGPT Prompt Generator

A decision-ready comparison brief, scoped tightly enough that the model does not flail.

You are a [INDUSTRY] strategy analyst.

Write a competitive teardown of [COMPETITOR A], [COMPETITOR B] and [COMPETITOR C] for [AUDIENCE, e.g. our exec team] who must decide [DECISION].

Context: we sell [PRODUCT] to [CUSTOMER SEGMENT]. Our current edge is [EDGE]; our known weakness is [WEAKNESS].

Acceptance criteria:
- One paragraph per rival covering positioning, pricing posture and the single reason we win or lose against them.
- A closing table ranking all three by threat level with a one-line justification each.
- Every claim names the evidence behind it or is marked as an assumption.

Format: Markdown, under [WORD COUNT] words, no jargon.

Reviews that return prioritised, actionable findings instead of generic advice.

You are a senior [LANGUAGE] engineer reviewing code for a [CONTEXT, e.g. production payments service].

Review the code below for correctness bugs, security issues and maintainability problems, in that order of priority.

Stack: [LANGUAGE + VERSION], [FRAMEWORK], [RUNTIME]. Conventions: [LINTER/STYLE GUIDE].

Acceptance criteria:
- Every finding names the file and line, the concrete failure it causes, and a fix.
- Findings are ordered most severe first.
- Say explicitly if you find nothing in a category rather than inventing filler.

Format: numbered list, one finding each, code snippets in fenced blocks.

Code:
[PASTE CODE]

Structured writing with a defined reader and a real angle.

You are a [NICHE] writer with hands-on experience in [DOMAIN].

Write a [WORD COUNT]-word article titled "[WORKING TITLE]" for [AUDIENCE], who already know [ASSUMED KNOWLEDGE] but not [THE GAP].

Angle: [THE SPECIFIC ARGUMENT OR INSIGHT]. Do not write a neutral overview.

Acceptance criteria:
- Opens with a concrete scenario, not a definition.
- Each section makes one claim and supports it with a specific example.
- Ends with something the reader can do this week.

Format: Markdown with H2 sections, [NUMBER] sections, [TONE] tone.

Analysis where the decision criteria are stated up front so trade-offs are made correctly.

You are a data analyst supporting [TEAM].

Analyse the dataset below and answer: [THE SPECIFIC QUESTION].

Context: this feeds [DECISION]. We care most about [PRIMARY METRIC]; [SECONDARY METRIC] is a tiebreak. Known data caveats: [CAVEATS].

Acceptance criteria:
- Lead with the answer in one sentence, then the supporting analysis.
- Quantify every claim and state the confidence.
- Flag anything the data cannot support rather than estimating around it.

Format: executive summary, then a findings table, then caveats.

Data:
[PASTE DATA]

Pattern-matching work — the one place few-shot examples still earn their keep.

You are a customer support lead for [PRODUCT].

Rewrite the draft reply below to match our voice: [VOICE, e.g. warm, direct, never defensive].

Rules: acknowledge the issue first, give the concrete next step with a timeframe, never blame the customer, never promise a date engineering has not committed to.

Example of a good reply:
"[PASTE ONE REAL EXAMPLE THAT MATCHES YOUR VOICE]"

Format: plain email text under [WORD COUNT] words, no subject line.

Draft:
[PASTE DRAFT]

Turns a transcript into something assignable, with explicit handling for ambiguity.

You are a chief of staff.

Turn the transcript below into a decisions-and-actions record for [AUDIENCE].

Acceptance criteria:
- Separate DECISIONS (already made) from ACTIONS (still to do) from OPEN QUESTIONS.
- Every action names an owner and a due date drawn from the transcript.
- Where an owner or date was never stated, write "unassigned" rather than guessing.

Format: three Markdown sections, bullet points, no preamble.

Transcript:
[PASTE TRANSCRIPT]