Verified against Gemini · 2026-06-12
Synthesize hundreds of pages of scattered documents into one decision memo
A long-context prompt for Gemini's 1M+ token window that reads every document dropped into the chat as one connected body of evidence and returns a single decision memo instead of a stack of disconnected summaries.
The prompt
Ready to copy — highlighted parts are example details you can swap.
I'm giving you 7 documents in this conversation (pasted or uploaded) about whether to renew or switch our warehouse management vendor. Read all of them as a single connected body of evidence before you write anything — don't summarize them one at a time. CONTEXT Decision to be made: whether to renew or switch our warehouse management vendor Audience for the memo: the ops leadership team, none of whom read the underlying contracts INSTRUCTIONS 1. Read every document fully before drafting. If any document appears truncated or you could not read all of it, say so explicitly before continuing — don't silently summarize a partial read as if it were complete. 2. Build one shared fact base across all documents. Note the source (file name or document title) next to every fact you use. 3. Flag every place where two or more documents disagree, contradict each other, or give different numbers for the same thing. Don't quietly pick one number and drop the other. 4. Group findings under: Cost comparison, Contract risk, Integration effort, Vendor track record 5. End with a "What's still unclear" section listing questions the documents don't answer — don't invent an answer to fill the gap. OUTPUT FORMAT A memo, not a list of summaries — no more than two pages, written for the ops leadership team, none of whom read the underlying contracts, with inline source citations like (Source: filename, page/section if visible).
Customize the highlighted detailsoptional — the prompt above already works
Why this works
Gemini's long-context window can hold every source document in the same context at once, which is the whole point of using it here — a synthesis built from documents summarized one at a time in separate turns loses the ability to notice that Document 3's number contradicts Document 5's, because by the time it reads Document 5 it never re-reads Document 3 closely. Reading everything in one pass keeps that cross-document comparison possible. The explicit instruction to flag contradictions instead of resolving them matters because long-context models have a documented tendency to quietly average or pick between conflicting numbers rather than surface the conflict, which hides exactly the information a decision-maker needs. The partial-read disclosure requirement guards against the failure mode where a very long or malformed upload gets truncated silently and the model summarizes what it saw as if it were the whole document.
What you get back
DECISION MEMO — Warehouse Vendor Renewal Cost comparison: Current vendor quotes $4.20/unit (Source: renewal_quote.pdf), incumbent contract cites $3.95/unit for the same tier (Source: original_msa.pdf, §4.2) — these two documents disagree on the current effective rate; worth confirming which is live before deciding. Contract risk: 90-day termination notice required (Source: original_msa.pdf, §9)... What's still unclear: None of the seven documents state the actual current inventory volume, so the cost comparison above is per-unit only, not a total spend projection.
Verified against
Gemini Gemini 3 Pro · 2026-06-12
Gemini Gemini 3 Flash · 2026-06-19
Changelog
- 2026-06-12 — Initial publish, verified against Gemini 3 Pro with a 7-document synthesis.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
