Who is Trusted-source brief for?
Students who have to answer a research question in a page and defend where every number came from.
Tools in this guide
0%0/5 steps
Student researcher · 75–90 minutes
You can hand in eight citations and still not be able to say which of them you opened.
This runbook builds the source trail first, so the brief survives the question every marker asks: where does this come from?
Trusted-source brief
Trusted-source brief
One page: question and exclusion rule, a table of 6–8 sources you opened with type and checked date, three claims with a strongest and weakest source each, and a methods note.
Claim 2: recall drops when the tool writes the summary and nobody rereads it. Strongest: Chen 2024, n=94, one semester, self-reported study time. Weakest: a vendor blog I kept as a counter-example. Not covered: anything past one semester.
Workbench downloads
Free templates for the brief. Fill them before you trust an AI answer — no account required.
Write one question you could answer in a page, one boundary (years, country, population), and one claim you refuse to make. Perplexity may list what falls outside the boundary. It does not get to pick the question.
Three lines at the top of the brief: question, boundary, exclusion rule.
My assignment: [assignment]. Propose three narrower questions I could answer in one page. For each, name the evidence that would fall outside it. Do not choose one for me and do not list sources yet.
Pitfall
Opening a search box first. Ten minutes later the assignment has quietly become whatever the results page was good at.Verification
You can say why a plausible paper is out of scope without reopening the chat.Search Semantic Scholar for papers and reviews, and follow the citations of the two that fit best. Then run one Perplexity pass for context papers miss: a new regulation, a national report, the institution running the pilot. Stop at ten.
Ten rows: title, DOI or URL, source type, one line on why it might belong.
For [question], list reports, agencies, or news from the last 24 months that academic search would miss. Give me openable URLs and the publisher of each. This is a lead list, not my bibliography.
Pitfall
Saving forty hits because the searching felt productive. You will keep six, and you will not have read the other thirty-four.Verification
Every row already names a source type and a reason to open it.Open all ten. Keep 6–8. Give each survivor a type (primary paper, review, policy, news, vendor, forum), the date you checked it, and one line in your own words on what it actually measured. Delete the rest instead of parking them in a maybe folder.
A source table where every row was opened by you and carries a date.
Here are my source titles and my one-line notes: [notes]. Flag every note that repeats the abstract instead of saying what the study measured and on whom. Do not add sources and do not rewrite my notes.
Pitfall
Pasting the abstract into the note field. Two days later you cannot tell whether you read the paper or the summary of it.Verification
No row is missing a type, a checked date, or a sentence in your own words.Write three claims. Under each, name the strongest source, the weakest one you kept, and the sentence that says what the evidence does not cover. Load only the saved sources into NotebookLM and ask where they disagree.
Three claims, each with a strongest source, a weakest source, and a caveat.
Using only the sources in this notebook, list every place two of them disagree. Quote the sentence from each source that creates the conflict. Do not resolve the conflict and do not write my claims.
Pitfall
Accepting a summary that merges two studies with different samples into one confident finding.Verification
Every claim has a caveat a marker can check against the source table.Write the page in your own words. Close with a methods note: which tools you used, for which step, and what you checked by opening the source. Where the assignment asks for your reasoning, generated prose does not count as reasoning.
The finished one-page brief with a methods note.
Here is my source table and claim list: [paste]. List rows with no checked date, claims with no caveat, and claims whose only support is a vendor page. Do not rewrite my sentences and do not make the tone more academic.
Pitfall
Handing in prose that sounds more certain than the six sources behind it. A marker who opens two of them will notice.Verification
Someone holding only your brief can rebuild the source trail.You should now have: Trusted-source brief
- Every source in the table was opened, typed, and dated by you.
- Every claim names a strongest source, a weakest source, and what the evidence does not cover.
- The methods note says which tool did which step and what you verified by hand.
- For how well answer engines cite, see the Citation Fidelity Lab protocol at /benchmarks/citation-fidelity once results are published.
PrivacyUnpublished participant data, private interview transcripts, sealed exams, and institution-confidential PDFs stay out of cloud tools. Work from public sources and readings you are licensed to process, or keep the file local. Disclose every tool that shaped the brief, including the ones that only helped you search.
Questions this playbook answers
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
What do you finish Trusted-source brief with?
One page: question and exclusion rule, a table of 6–8 sources you opened with type and checked date, three claims with a strongest and weakest source each, and a methods note.
How long does Trusted-source brief take?
75–90 minutes
What goes wrong in Trusted-source brief?
Each step names the mistake you are about to make:
- Opening a search box first. Ten minutes later the assignment has quietly become whatever the results page was good at.
- Saving forty hits because the searching felt productive. You will keep six, and you will not have read the other thirty-four.
- Pasting the abstract into the note field. Two days later you cannot tell whether you read the paper or the summary of it.
- Accepting a summary that merges two studies with different samples into one confident finding.
- Handing in prose that sounds more certain than the six sources behind it. A marker who opens two of them will notice.