The 4PM Friday Pattern
Every team I’ve worked with has the same Friday afternoon pattern. Someone runs an AI prompt. The output is impressive. They forward it to a colleague or stakeholder. Two hours later they get a reply: “This is good but I need it as a 1-page summary, not a 5-page brief. Also, can you add a recommendation? Also, who is this for?”
The AI output was correct. It just wasn’t useful in the form it arrived.
This is not a model problem. This is a Handoff problem.
What Handoff (READY) Actually Means
Handoff is the H in ORCHESTRATE — the systematic prompting framework that breaks down what a great prompt contains. Most prompting advice focuses on the first three components: Objective, Role, and Context. Those matter. They drive about 80% of the quality improvement.
But the next 15% comes from four enhancement components, and Handoff is the one most people skip.
The acronym for Handoff is READY:
- R — Recipient: Who is actually going to read or use this output?
- E — Exact format: What is the deliverable physically? (PDF, Slack message, slide, email body, code review comment, voice memo script)
- A — Application: What will the recipient do with this output?
- D — Decisions enabled: What specific choice does this output unlock?
- Y — Yes-criteria: What does “good enough to act on” look like?
When all five are explicit in the prompt, AI stops producing “comprehensive overviews” and starts producing artifacts a human can immediately use.
A Worked Example
Consider this generic prompt:
Summarize the attached customer interview transcript and pull out the key insights.
Here is the same prompt with Handoff specified:
Summarize the attached customer interview transcript.
Recipient: Our VP of Product, who has 5 minutes between meetings.
Exact format: A Slack message, max 200 words, with the headline as the first line in bold.
Application: She’ll forward this to the design team to inform a sprint planning conversation tomorrow.
Decisions enabled: Whether to add the requested feature to next sprint or defer to backlog.
Yes-criteria: I should be able to forward this without editing if it (a) names the customer’s actual job-to-be-done in their words, (b) flags any blocker that would make our current solution unusable, and (c) gives a clear “ship it / defer” recommendation.
The first prompt produces a wall of text and a list of insights. The second prompt produces a forwardable Slack message that drives a specific decision.
The difference is not the AI. The difference is that the second prompt actually told the AI what success looks like.
The Test: Can You Forward It Without Editing?
The single best heuristic for whether your Handoff specification is tight enough is this: can the recipient act on the output without asking you a clarifying question?
If the answer is no, you skipped a Handoff field.
- If they ask “who is this for?” — you skipped Recipient.
- If they ask “can you put this in a deck?” — you skipped Exact format.
- If they ask “what should I do with this?” — you skipped Application.
- If they ask “what are you recommending?” — you skipped Decisions enabled.
- If they ask “is this final?” — you skipped Yes-criteria.
Each clarifying question represents 5–30 minutes of round-trip rework. Across a team of 50 knowledge workers running 10 AI prompts a day, the math gets ugly fast.
Why Yes-Criteria Is the Cheat Code
Of the five Handoff fields, Yes-criteria is the one most people miss even after they hear about the framework.
Yes-criteria is the contract. It tells the AI (and you, when you review the output) what “ready to ship” actually means. It is not “make it good.” It is “the headline must reference the customer’s actual words, the recommendation must be ship-or-defer, and there must be no more than three bullet points.”
Yes-criteria is also the cheat code for self-review. Once it’s in the prompt, you can ask the AI to grade its own output against the criteria before you read it. Half the time it catches its own gaps and rewrites without you needing to.
Three Templates You Can Steal
Executive summary template:
Recipient: [Name + role + minutes available] Exact format: [Word count + structure] Application: [Specific meeting or decision] Decisions enabled: [The choice this unlocks] Yes-criteria: [3-5 specific quality bars]
Customer email template:
Recipient: [Customer name + relationship stage + last interaction] Exact format: [Email body, subject line, signature block] Application: [What you want them to do next] Decisions enabled: [Reply Y/N? Calendar invite? Forward internally?] Yes-criteria: [Tone, length, specific phrases to use or avoid]
Code review template:
Recipient: [Author + their experience level + their authorial intent] Exact format: [GitHub PR comment, inline annotations, summary block] Application: [Will they refactor today, file follow-up tickets, or defer?] Decisions enabled: [Approve / request changes / block] Yes-criteria: [Categories to comment on, severity threshold, no nitpicks]
The Habit Shift
Adopting Handoff is not about memorizing READY. It’s about a habit shift: before you hit send on a prompt, spend 90 seconds writing two or three sentences about who receives the output and what they need to do with it.
Most people resist this because it feels like overkill for a quick task. But “quick task” is exactly when the rework hurts most — you save five minutes on the prompt and lose forty-five on the back-and-forth.
The teams I’ve watched adopt Handoff systematically report the same thing: their AI workflow stops feeling like a gamble and starts feeling like delegation.
That’s the actual goal. Not impressive AI. Useful AI.
This article is adapted from a LinkedIn series on the ORCHESTRATE method.
