Workflow

Advanced

Deep research brief pack

Advanced

Frames questions, synthesizes evidence (where available), and lists what still requires primary research.

Overview


"Market scan" requests often return undigested links or confident guesses. This pattern forces decision framing, confidence tiers, and explicit unknowns—appropriate when benchmarks, regulation, or vendor claims need disciplined scrutiny.

Who should use it
Strategy, partnerships, and senior ICs packaging research for investment, procurement, or GTM choices—especially under information boundaries.

Use it for

  1. Vendor or product landscapes where short lists must map to your constraints, not generic quadrants.

  2. Regulatory or data-handling touchpoints where misstating nuance is worse than saying unknown.

  3. Benchmark requests where public data is thin and stakeholder optimism needs a guardrail.

Inputs

  1. The decision the research should inform—and what "good enough" evidence looks like.

  2. Geography, segment, and technology boundaries.

  3. Sources you may use (paid, internal, partner) and topics or files that must never leave trusted systems.

Context


Specify citation expectations and forbid fabricating statistics. Where Deep Research or connectors are used, remind readers that synthesis is not primary data—pair model output with human verification, especially for regulated contexts.

Prompt Skeleton

Create a research brief. Decision question: ... Boundaries: ... Task: 1) Clarify the decision frame 2) Key unknowns (questions) 3) Evidence map (what we know / partial / unknown) 4) Scenario commentary (2-3 futures) clearly labeled speculative 5) Recommended next research steps + owners 6) Governance notes: data you should NOT paste externally If data is missing: say so.

Copy

Review checklist

  1. Sensitive documents were not pasted into environments your policy forbids.

  2. Claims are tied to sources or flagged as uncited inference.

  3. Recommendations scale with evidence quality—no false precision.

Common mistakes

  1. Treating model synthesis or web snippets as authoritative primary research.

  2. Sharing regulated files with the wrong workspace or connector scope.