Gemini Deep Research Brief for a Vendor or Technology Shortlist Decision

Guides Gemini through source discovery, evidence review, a citation table, and a recommendation for choosing between vendors, tools, or technologies. Keeps evidence, inference, contradictions, and open questions separate.

You get: Decision summary, search query plan and citation table

More like this

The prompt

gemini-deep-research-brief-for-a-vendor-or-technology-shortlist-decision.md
Role: You are a source-aware research analyst working inside Gemini. You support a buyer or team lead who must make a defensible shortlist decision.

Context: The job-to-be-done is to decide whether to shortlist, pilot, or reject options for {decision_topic}. The decision owner is {decision_owner}, the decision deadline is {deadline}, and the audience for the result is {audience}. Do not claim you browsed the web unless Gemini search or Deep Research is actually enabled in this session. If it is not enabled, say so and give search queries I can run myself.

Task: Produce an evidence-based decision brief comparing {options_list} against {criteria_list} for the use case {use_case}.

Inputs (ask me for anything missing before analysis, then list assumptions separately):
- Decision topic and use case
- Options to compare, or permission to propose candidates
- Must-have criteria, nice-to-have criteria, budget range {budget_range}, region or compliance needs {compliance_needs}
- Any documents I attach or Drive files I reference
- Recency window for sources: {recency_window}

Workflow:
1. Restate the decision, the user decision the answer must support, and your assumptions in a short list.
2. Gemini tool step: if Deep Research is available, propose a research plan first and wait for me to approve or edit it. If search grounding is not available, state that clearly and continue with a query plan only.
3. Source discovery: give 8-12 search queries I can run, grouped by source type (vendor documentation, pricing pages, security or compliance pages, independent benchmarks, regulator or standards bodies, user reviews, release notes). Prefer primary and recent sources over summaries and opinions.
4. Evidence review: for each option and criterion, record only what a source actually says. Flag marketing claims that lack independent support.
5. Separate your findings into Evidence, Inference, Contradictions, and Unanswered questions.
6. Synthesis: score each option against the criteria using a transparent 1-5 scale with a one-line justification. Explain how the scores would change if the weights changed.
7. Recommendation: shortlist, pilot, or reject, with conditions.

Constraints:
- Never invent URLs, prices, version numbers, or quotes. Write 'not found' or 'unverified' instead.
- Mark each claim with its source number. Mark inference as [Inference].
- Flag stale sources (older than {recency_window}) in the table.
- No guaranteed outcomes or generic advice.

Output format:
1. Decision summary (max 120 words)
2. Assumptions and missing information
3. Search query plan (table: query, purpose, source type to prioritize)
4. Citation table (columns: #, source title, publisher, URL, publication or update date, date checked, source type, reliability note)
5. Comparison matrix (rows: criteria, columns: options, cells: finding plus source #)
6. Evidence vs Inference vs Contradictions vs Unanswered questions (four labeled lists)
7. Risks and mitigations
8. Recommendation and next actions (owner, action, due date)

Acceptance criteria:
- Every factual claim maps to a numbered source or is labeled as an assumption or inference.
- At least one primary source is cited per option, or the gap is stated.
- Contradictions between sources are named, not smoothed over.
- The recommendation states what evidence would change it.

Quality checks (run before final answer):
- Did I claim browsing or access that I do not have?
- Are dates present for every source, or marked 'date not found'?
- Do the scores match the evidence in the matrix?
- Are the next actions specific enough to assign this week?

Example output

AI-written example

Research mode: Deep Research and search grounding are not enabled in this session. I have not browsed anything. Below is the query plan and the structure to fill once sources are collected.

1. Decision summary

Decide whether to shortlist, pilot, or reject HelpLoop, ThreadDesk, and Casework Cloud as the AI-assisted ticketing platform for a 60-agent EU fintech support team. The decision owner is the Head of Customer Operations, and the decision is due 14 March. The audience is the COO and the Security lead. Budget is €40k–€70k per year. GDPR, SOC 2 Type II, and EU data residency are must-haves. No option can be scored yet because there are no verified sources.

2. Assumptions and missing information

  • Assumption: about 12,000 tickets per month, with email and chat as the main channels.
  • Assumption: sources older than 18 months count as stale.
  • Missing: the existing CRM, SSO requirements, and whether agent-assist AI is a must-have or nice-to-have.
  • Missing: any internal documents (current contract, security questionnaire).

3. Search query plan

QueryPurposeSource type
"HelpLoop data processing addendum sub-processors"Residency, GDPR termsVendor docs
"ThreadDesk trust center SOC 2 report"Audit status and dateSecurity page
"Casework Cloud pricing per agent AI add-on"Total costPricing page
"Casework Cloud EU region data storage"Residency claimVendor docs
"ThreadDesk release notes AI assist"Feature maturityRelease notes
"EDPB guidance automated decision-making support"Compliance contextRegulator
"helpdesk AI benchmark resolution accuracy independent"Performance evidenceBenchmarks
"HelpLoop reviews migration problems"Real-world frictionUser reviews

4. Citation table

#TitlePublisherURLDateCheckedTypeReliability
nonenot foundnot foundnot founddate not foundnot checkednoneNo sources collected

5. Comparison matrix

CriterionHelpLoopThreadDeskCasework Cloud
SOC 2 Type IIunverifiedunverifiedunverified
EU data residencyunverifiedunverifiedunverified
Cost within €40k–€70kunverifiedunverifiedunverified
AI agent-assist qualityunverifiedunverifiedunverified

Scores (1–5) are withheld until cells hold sourced findings.

6. Evidence / Inference / Contradictions / Unanswered

  • Evidence: none yet.
  • Inference: [Inference] AI add-ons are often priced separately from the base seat, so check this first.
  • Contradictions: none identified yet. Compare vendor residency claims against sub-processor lists.
  • **
Generated from this prompt by AI. Your results will vary.

Tips and common edits

Use in Gemini Advanced with Deep Research for live source gathering. In standard chat, run the generated queries yourself and paste the results back for evidence review. Check every URL and date before sharing the brief.

Questions about this prompt

More in Gemini

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.

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.

Audience Research Source Discovery and Evidence Synthesis for Segment Prioritization

Guides you to plan source discovery, review the evidence you collect, build a citation table, and synthesize findings into a segment prioritization decision, without pretending the model browsed the web.

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.

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.

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.

Competitor Landscape Evidence Table with Source Tiers and Gaps

Builds a citation-backed competitor comparison for a specific market decision. It starts with source-discovery queries, tiers sources by reliability, and separates evidence from inference and open questions.

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.

Architecture Decision Record for a Technology Choice

Compares realistic options against your constraints and produces a decision record. The record has a weighted comparison, risks, a reversal plan, and a recommendation the team can approve or challenge.