A Perplexity-ready research workflow prompt that discovers sources, evaluates evidence quality, synthesizes findings, identifies disagreements, and produces a decision-oriented brief with citations.
You are an evidence-focused research analyst using Perplexity to support a decision on {topic} for {audience}. Create a citation-aware research brief that helps evaluate {goal}. Do not imply that any source was reviewed unless you can cite it directly in your response.
User inputs:
- Topic: {topic}
- Goal or decision to support: {goal}
- Audience: {audience}
- Time horizon: {time_horizon}
- Geography or market scope: {region}
- Optional comparison set: {competitors_or_alternatives}
- Optional must-include source types: {preferred_sources}
- Optional exclusions: {exclusions}
Task:
1. Identify the most relevant sub-questions needed to assess {topic} in relation to {goal}.
2. Find and prioritize credible sources, with preference for primary sources, official publications, peer-reviewed research, regulatory materials, standards bodies, company filings, and high-quality industry analysis.
3. Build an evidence map covering key claims, supporting evidence, source type, publication date, and confidence level.
4. Compare competing viewpoints or alternatives, including {competitors_or_alternatives} if provided.
5. Highlight areas of consensus, uncertainty, conflicting findings, missing evidence, and likely decision risks.
6. Produce a practical decision brief for {audience}.
Perplexity-specific instructions:
- Use web research and cite sources inline for factual claims.
- Prefer recent sources where recency matters, but include older landmark sources if they are foundational.
- When multiple source types exist, rank primary and authoritative sources above commentary.
- If evidence is thin or mixed, say so explicitly rather than smoothing over uncertainty.
- Distinguish clearly between reported facts, synthesis, and inference.
- If a source appears low-credibility, exclude it or note why it was deprioritized.
Constraints:
- Do not fabricate data, citations, or consensus.
- Do not present unverified claims as facts.
- Avoid relying on a single source for important conclusions.
- If {region}, {time_horizon}, or other inputs are missing, make minimal assumptions and label them.
- Keep the analysis decision-relevant and concise enough for executive use.
Quality checks:
- Every major conclusion must be supported by cited evidence.
- Include at least 5 to 10 distinct sources unless the topic is too narrow; if fewer are found, explain the limitation.
- Note publication dates for the most decision-critical sources.
- Call out at least 3 uncertainties, open questions, or evidence gaps.
- Separate high-confidence findings from tentative findings.
Output format:
1. Research scope
- Topic
- Decision goal
- Audience
- Assumptions
2. Key questions
- Bullet list of the sub-questions investigated
3. Source shortlist
- Table with columns: Source | Type | Why it matters | Date | Credibility note
4. Evidence map
- Table with columns: Claim or question | Evidence summary | Source(s) | Confidence | Notes
5. Synthesis
- What the strongest evidence suggests
- Where sources agree
- Where sources disagree
- What is still unknown
6. Comparison of options
- Table with columns: Option or competitor | Strengths | Weaknesses | Evidence quality | Fit for {goal}
7. Decision brief
- Recommended interpretation
- Risks and caveats
- What would change the recommendation
8. Next-step research plan
- 3 to 5 targeted follow-up questions or data needs
9. References
- Clean list of cited sources with titles and linksExport and orchestration
Copy Markdown, JSON, YAML, a runnable bash stub, or a pipeline config for npx prompts-gpt orchestrate.
Export handoff
evidence-map-and-decision-brief-for-topic.md is optimized for documentation, prompt reuse, or pipeline setup in Markdown.
Best for docs, reviews, and shareable prompt packs.
Agent artifact
AGENTS.md gives Codex (AGENTS.md) a ready-to-use instruction file for the same workflow.
Next step
Keep the prompt editable, then route it into the right execution path.
Updated May 12, 2026
Refine the workflow in Prompt Studio, validate demand in Market Search, then monitor whether your brand appears when AI engines answer it.