The Fable 5 One-Million-Token Diligence Memo (Copy, Paste, Done in 20 Minutes)

Reading a stack of long documents to find the handful of things that actually matter is the most delegatable work an executive does, and the hardest to delegate well. You need the judgment to know what is a real risk, but the task itself is mechanical: read everything, cross-reference it, surface the exceptions. Until recently, AI could not help much because it could not hold the whole stack in view at once.

Fable 5 changes that. With a one-million-token window, it reads a full data room or a contract set in a single pass and answers across all of it. The prompt below is built for that. It tells Fable 5 to plan first, pause for your approval, then produce a one-page memo where every claim points to the exact source line so you can verify it in seconds.

A half-day read, done as a supervised twenty-minute session. Here is the whole thing.

This is a PromptHacker Premium deep dive.

The full copy-ready prompt, the approval checkpoint, and the reuse cases are below.

Step 1: Load the Documents

In Claude, select Fable 5 in the model picker, then attach the full set of documents to a single chat: contracts, financial statements, vendor reports, prior correspondence, whatever the decision depends on. Because the window is a million tokens, you do not need to split them. Attach them all.

Step 2: Paste This Prompt

You are my diligence analyst. I have attached every document for a decision I am evaluating. Work in two stages and stop after stage one. STAGE ONE: Plan only. Tell me (a) what each document is, (b) the questions you will answer to assess risk and opportunity, and (c) anything you notice is missing or inconsistent at first glance. Do not analyze yet. Wait for my approval. STAGE TWO (only after I say "approved"): Produce a one-page memo with these sections: 1. The five most material risks, ranked. For each, cite the exact document and line or clause, and label your confidence as High, Medium, or Low. 2. Key obligations and deadlines (renewal, termination, payment, exclusivity), each with a source citation. 3. The three open questions I should resolve before deciding, and who can answer each. 4. A two-sentence plain-language bottom line for a non-specialist reader. Rules: cite a source for every factual claim. If something is ambiguous, say so rather than guessing. Do not pad. If a section has nothing material, write "none found" and move on.

Step 3: Approve the Plan, Then Read the Memo

The two-stage structure is the whole trick. Stage one shows you how Fable 5 understands the material before it commits to conclusions, which is your chance to catch a misread document or add context it is missing. Once you reply "approved," it produces the memo. Because every claim cites a source line, your review is fast: you are not re-reading the data room, you are spot-checking the five citations that carry the most weight.

Why it works: the citation requirement turns the model from something you have to trust into something you can check. Verification is the bottleneck, and a source line per claim removes it.

Three Other Jobs to Reuse It On

The pattern is not just for acquisitions. Swap the documents and it handles: a vendor renewal , where you load the current contract plus the proposed terms and ask what changed and what it costs you; a lease or office decision , where you load the agreement and comparable terms; and an insurance or policy review , where you load the policy set and ask what is actually covered. Same prompt, same two-stage discipline.

Action Steps Summary

• Select Fable 5 and attach everything. The full document set in one chat, no splitting.
• Paste the two-stage prompt. It plans first and waits.
• Review the plan before approving. Catch misreads and add missing context.
• Spot-check the top five citations. Verify the claims that matter, not the whole stack.
• Save it as your diligence template. Reuse it on renewals, leases, and policy reviews.