Parallel Research Fan-Out and Fan-In with Citation Table and Decision Memo

Splits a research question across parallel research agents, defines source discovery and evidence review steps, merges findings into a citation table, and produces a decision memo without claiming live browsing.

You get: Every factual claim in the synthesis maps to a source ID…, retrieval status is stated honestly for every source and conflicts between sources are shown

More like this

The prompt

parallel-research-fan-out-and-fan-in-with-citation-table-and-decision-memo.md
ROLE
You are a research pipeline coordinator running parallel sub-agents (or parallel sections) and a final synthesis agent.

CONTEXT
Decision to support: {decision}. Topic: {topic}. Sub-questions: {sub_questions}. Preferred source types: {source_types}. Date range that matters: {date_range}. Risk tolerance: {risk_tolerance}. Platform: {tool_platform}.
Job to be done: help me choose between options with traceable evidence and know how confident to be.

TASK
Fan out the topic into independent research lanes, specify how sources should be found and reviewed, merge the results into a citation table, then write a decision memo.

CONSTRAINTS
- Do not claim you browsed the web, opened links, or checked live data unless this platform actually returned retrieved sources in this conversation. If no retrieval is available, produce search queries, source-type targets, and a review checklist, then ask me to paste sources.
- Never invent URLs, quotes, statistics, authors, or dates. Mark anything from model memory as 'unverified recall'.
- Keep lanes independent: no lane may rely on another lane's conclusions.
- Separate stated facts, assumptions, and open questions.
- Platform notes: in Perplexity or ChatGPT with search, include the URL and retrieval date for each source you actually retrieved. In Claude without search, work only from text I paste and cite the paste label (S1, S2...). In Gemini Deep Research, ask it to return a source list first, then synthesize.

INPUTS
Check that I provided decision, topic, sub_questions, source_types, date_range, risk_tolerance, tool_platform. Ask up to 5 questions for anything missing, such as the options being compared or the minimum evidence needed.

WORKFLOW
1. Define the decision criteria (3-6) and the options being compared.
2. Create one lane per sub-question. For each lane give: lane goal, 3 search queries, preferred source types, inclusion/exclusion rules, and what would change the decision.
3. Run each lane (or request pasted sources). Extract claims with source labels.
4. Evidence review: grade each source for relevance, recency, independence, and method (High / Medium / Low) with a one-line reason.
5. Fan-in: merge into the citation table, mark conflicts between lanes, and explain which evidence you weight more and why.
6. Write the decision memo with a recommendation, confidence level, and what would flip it.

OUTPUT FORMAT
A. Decision framing: options, criteria, assumptions
B. Lane plan table: Lane | Question | Queries | Source types | Exclusions
C. Citation table: ID | Source and URL (or 'pasted text') | Date | Claim | Evidence grade | Lane | Retrieved by tool? (Y/N)
D. Conflicts and gaps list
E. Synthesis: 5-8 bullets, each tagged with source IDs
F. Decision memo (under 300 words): recommendation, confidence (Low/Med/High), risks, missing information
G. Next actions: 3-5 steps with owner and effort

ACCEPTANCE CRITERIA
- Every factual claim in the synthesis maps to a source ID or is labeled unverified.
- Retrieval status is stated honestly for every source.
- Conflicts between sources are shown, not smoothed over.
- The memo states confidence and what evidence would change the recommendation.

QUALITY CHECKS
Before finalizing: confirm no fabricated links or quotes; confirm each lane has at least one inclusion rule and one exclusion rule; confirm the memo does not use evidence absent from the citation table; confirm confidence matches the evidence grades. List any check that failed.

Example output

AI-written example

Status: No retrieval is available in this conversation and no sources have been pasted. Below are the lane plan, queries, and review checklist. The citation table is empty until you paste sources (label them S1, S2…).

A. Decision framing

  • Options: (1) managed Postgres from a major cloud provider, (2) self-hosted Postgres on VMs.
  • Criteria: 3-year total cost, reliability and recovery time, security and compliance (SOC 2 planned), ops workload for a 12-person team with no DBA, lock-in and migration effort.
  • Assumptions (unconfirmed): about 400 GB of data, one primary plus replicas, low risk tolerance, 2022–2025 sources preferred.

B. Lane plan

LaneQuestionQueriesSource typesExclusions
1 CostWhat is the real 3-year cost gap?"managed vs self-hosted PostgreSQL total cost 2024"; "RDS pricing vs EC2 Postgres TCO"; "database ops engineer hours per month"Vendor pricing pages, engineering blogs with numbersPre-2022 pricing; posts without figures
2 ReliabilityHow do outage and recovery outcomes compare?"managed Postgres incident postmortem"; "self-hosted Postgres failover RTO"; "provider SLA Postgres uptime"Postmortems, SLAs, peer-reviewed or conference talksMarketing claims with no method
3 ComplianceWhich option is easier to audit for SOC 2?"SOC 2 managed database shared responsibility"; "Postgres encryption backups audit evidence"; "self-hosted database compliance burden"Provider compliance docs, auditor guidanceVendor-neutral claims with no citation

Each lane would change the decision if: cost gap is under 15%, a managed-service outage hit a comparable workload, or auditors require controls only self-hosting allows.

C. Citation table

IDSource and URLDateClaimGradeLaneRetrieved by tool?
nonenonenonenonenonenoneN

Review checklist per source: relevance to our scale, published within the date range, independent of the vendor, method stated (High, Medium, or Low, with a one-line reason).

D. Conflicts and gaps

  • No evidence yet in any lane.
  • Missing inputs: actual data size, current monthly spend, SLA targets.

E. Synthesis

  • Managed services typically reduce ops workload for small teams (unverified recall).
  • Self-hosting can be cheaper at large scale but needs dedicated staff (unverified recall).
  • Nothing else can be
Generated from this prompt by AI. Your results will vary.

Tips and common edits

Use with a search-enabled tool when possible, and verify every link yourself before relying on it. Without search, paste source excerpts and label them S1, S2, and so on. Keep sub-questions to 3-5 so each lane stays focused.

Questions about this prompt

Research Agent Brief: Source Discovery, Evidence Review, and Citation Table

Directs an AI agent to plan source discovery, grade evidence, build a citation table, and support a decision without claiming live browsing occurred.

Competitor UX Pattern Research Plan with Evidence Table and Design Recommendation

Builds a source discovery plan, evidence log, citation table, and synthesis for comparing UX patterns across competitors, without claiming live browsing.

Competitor and Audience Evidence Review with Citation Table and Go/No-Go Brief

A research workflow that plans online source discovery, scores evidence, builds a citation table, and ends with a decision brief, without pretending live browsing happened.

Competitor Landscape Research Plan with Source Discovery and Citation Table

Builds a source discovery plan, evidence review, citation table, and decision-ready synthesis for a competitor or market question, without pretending live browsing occurred.

Competitor Landscape Scan with Source-Backed Citation Table

Runs a structured competitor scan that starts with search queries, ranks sources by reliability, and ends with a cited comparison table and a positioning recommendation.

Evidence Review Plan for a Classroom Teaching Strategy Decision

Guides source discovery, evidence review, a citation table, synthesis, and a recommendation on whether to adopt a teaching strategy. It does not claim live browsing and marks every unverified item for you to check.

Productivity Tool Selection: Source Discovery, Evidence Review, and Citation Table

A research workflow to choose between candidate productivity or automation tools. It plans online source discovery, structures evidence review in a citation table, and ends in a decision memo without claiming live browsing took place.

Software Vendor Due Diligence Brief with Cited Evidence Review

Guides a vendor shortlist review: finds primary sources like docs, security pages, and pricing, scores vendors on your criteria, and flags risks and open questions before purchase.

Strict JSON Handoff Schemas with Validation and Retry Rules for Agent Pipelines

Turns a pasted pipeline description into JSON schemas for every agent-to-agent handoff, plus validation checks and retry rules for bad output.

AI Paper Explainer with Evidence Table and Practical Takeaways

Turns a pasted AI paper abstract or text into a plain-language explanation of what is new, how it was tested, its limits and what to do with it.

Customer Voice Synthesis from Reviews and Forum Threads

Turns pasted reviews, support threads, and forum posts into coded themes, representative evidence, and a ranked list of product or messaging decisions.

Fact-Check Research Brief for an Explainer or Documentary Video

Builds a source discovery plan, evidence review, and citation table for a video's factual claims so a producer can decide which claims to keep, soften, or cut before scripting.