VerdictPal
Join
GuidesPlaybook

Trusted-source brief

A one-page brief with a question, source table, claim map, caveats, and privacy note.

Beginner
Start now
Playbook

Trusted-source brief

A one-page brief with a question, source table, claim map, caveats, and privacy note.

Steps5Time~85 minArtifactTrusted-source briefLevelbeginner
Before you start
  • Assignment topic chosen
  • Basic familiarity with web search
0%0/5 steps

Student researcher · 75–90 minutes

Most AI research fails because the sources are invisible.

This runbook makes the source trail visible before you write the answer.

Trusted-source brief

Trusted-source brief

One page: research question, 6–8 source rows, three claims, and named caveats.

Claim: AI note-taking tools can reduce recall if students skip retrieval practice. Caveat: most classroom evidence is small-N or self-reported.

Workbench downloads

Free templates for the brief. Fill them before you trust an AI answer — no account required.

Frame the question

10 minPerplexity

Write one answerable question, one scope boundary, and one claim you will not make.

Question block with boundary and exclusion rule.
Turn this assignment into one narrow research question. Name what evidence would be out of scope: [assignment].
Pitfall
Letting the tool decide the scope for you.
Verification
You can explain why a source is included or excluded.

Collect candidates

20 minSemantic Scholar

Search Semantic Scholar first, then use Perplexity to find current context or institutions you missed.

A 10-source candidate list.
Pitfall
Saving every search result before opening it.
Verification
Each candidate has title, URL/DOI, source type, and why it might matter.

Classify sources

20 minZotero

Open every candidate. Keep 6–8 and label each as primary paper, review, policy, news, vendor, or forum.

A source table that survives audit.
Pitfall
Treating search snippets as source reading.
Verification
No source row lacks a type, date checked, and one sentence of relevance.

Map claims

20 minNotebookLM

Draft three claims. For each: strongest source, weakest source, and what the evidence does not prove.

Three claim cards with caveats.
Pitfall
Using a grounded summary as if it resolved source disagreement.
Verification
Every claim has a caveat a marker could check.

Ship the brief

15–20 min

Write the one-page brief in your voice. Add a methods note that names tools used and what stayed human.

Final trusted-source brief.
Pitfall
Submitting polished generated prose instead of an auditable argument.
Verification
A reader can recreate the source trail from the brief alone.
Finish line

You should now have: Trusted-source brief.

  • Every source in the brief was opened, classified, and checked on a visible date.
  • Every claim has a source and a caveat; no generated paragraph stands alone.
  • The methods note names the tools used and what you verified manually.
  • For answer-engine citation checks, see the Citation Fidelity Lab protocol at /benchmarks/citation-fidelity once results are published.

PrivacyDo not upload unpublished participant data, private interview transcripts, or institution-confidential PDFs to cloud tools. Use public sources, licensed readings you are allowed to process, or local notes; disclose every AI tool that shaped the brief.