Research brief with source checking

You are a careful research analyst. Prepare a research brief on this question:
[RESEARCH QUESTION]

Purpose of the brief: [WHAT DECISION OR PIECE OF WORK IT SUPPORTS]
Time frame that matters: [e.g. "studies since 2022"]
Sources to use: [PASTE SOURCES, OR WRITE "search the web" IF YOUR ASSISTANT CAN BROWSE]

Work in this order:
1. Restate the question and list 3-5 sub-questions that together answer it.
2. For each sub-question, give the key findings. After every factual claim, cite the source in brackets.
3. Build a source table: Source | Type (peer-reviewed, official, news, opinion, vendor) | Date | Reliability (high/medium/low) | Why.
4. List points where sources disagree and explain the likely reason.
5. List what you could not find or verify. Do not fill gaps with assumptions.
6. End with a 5-sentence answer to the original question and a confidence level (high/medium/low).

Rules:
- Use only the sources provided or found. If you cannot access a source, say so.
- Never invent titles, authors, URLs or statistics.
- Mark any claim that relies on a single source as "single source".

Example (excerpt):
Sub-question 2: Does a four-day week reduce burnout?
- Several pilot programmes reported lower self-reported burnout after six months [Source 3]. Single source for the exact percentage.

| Source | Type | Date | Reliability | Why |
| Source 3 | Pilot report | 2023 | Medium | Large sample, but participating firms volunteered |

Not verified: long-term effects beyond 12 months; no source covered them.

When to use it

Use this prompt when you need to understand a topic well enough to make a decision or write about it: a market question, a health or policy topic, a technical comparison or background for a report. It is designed for questions where getting the facts wrong is costly, so it puts as much weight on source quality and gaps as on the answer itself.

How to customize it

• [RESEARCH QUESTION] should be specific. “Is remote work good?” is too vague; “What does research since 2020 say about remote work and junior employee productivity?” works.
• [WHAT DECISION OR PIECE OF WORK IT SUPPORTS] helps the model decide which findings are relevant.
• [TIME FRAME] stops outdated sources from dominating the brief.
• [SOURCES] is the most important choice. If your assistant cannot browse, paste articles, abstracts or reports directly. If it can browse, say so explicitly, and still check the links it returns.

Why it works

The prompt breaks one large question into sub-questions, a form of step-by-step decomposition similar to chain-of-thought prompting. Each step has a clear output, which makes the brief easier to check.

Requiring a citation after every claim and a reliability rating for every source forces the model to keep evidence and conclusions separate. The explicit list of unknowns and the “single source” label work against hallucination: instead of producing a confident but unsupported summary, the model is told that admitting a gap is part of the task. The final confidence level gives you a quick signal of how much further checking is needed.

Tips and variations

• Open every cited link yourself before relying on it. Models can misread or overstate what a source says, even when the source is real.
• For academic topics, paste abstracts and ask the model to note sample size and study design in the source table.
• Add “Write the brief for a reader with no background in [FIELD]” to change the level.
• Our article on how AI research search tools work explains why citations from search-based assistants still need checking.
• Run the same question in two different assistants and compare the source tables.