Customize each out-of-office reply.

What the AI will do

• Searches your correspondence, calendar, projects, and handover notes to understand what senders need and what you can truthfully tell them.
• Builds an automatic away responder that acknowledges people personally, uses confirmed commitments, and identifies every reply as AI-written.
• Tests realistic examples and actual delivery, then asks whether you are confident enough to turn it on.

What you can provide

• Your absence dates if they are not clear in your calendar, access to the relevant email and work context, and any handover details that are missing.

Where you'll take part

• Confirm the dates, what the AI may say, and who should receive replies. Review a few tests and decide whether to activate the responder.

What you'll get

A tested personalized out-of-office responder, turned on or scheduled for your absence after you approve, with a way to inspect and pause it. Draft replies alone are not the finished result.

Activity context

“I am out of the office” can become “Hi Kathy, I know I owe you that report.” The AI can acknowledge the person, the work you share, and your actual return plans—even when it cannot solve the request. Each reply makes clear that an AI wrote it while you were away.

For the user

Use the prompt to have the AI investigate your correspondence and email setup, build the responder, and run a few tests. You confirm the facts and sending rules, inspect example replies, and answer “Are you confident enough to turn this on?” The AI then activates the agreed system if you say yes.

Guidance for the AI

These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.

Start with the activity and the user's context

Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.

Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.

Check the capabilities needed for this work

Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.

When useful access is missing:

• Explain what you cannot reach and how that material would help.
• Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
• If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
• Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.

Ask focused questions and respect the answers

Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.

Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.

Keep evidence and independent contributions distinct

Distinguish evidence, inference, and missing information. Check your work against the source material.

Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:

• Give each agent a bounded brief and the context appropriate to its role.
• Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
• Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.

Produce, check, and revise the result

Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.

Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.

Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.

Your AI prompt

Make my out-of-office replies personal and automatic

Build and test an out-of-office system that responds to each sender's actual message and relationship with me. It should answer the need when it has a grounded, appropriate answer, or at least acknowledge the person and the unfinished business between us. The goal is a real automatic responder during my absence. Prepare it, run tests, ask the activation question below, and turn it on only after I say yes.

Establish my absence, correspondence, and actual email setup

Inspect the email account and its existing out-of-office settings, inbox rules, connected automation tools, and available read and send permissions before recommending changes. Read the current settings rather than assuming they are disabled or inadequate. Record their current state so the new workflow can be coordinated with them and restored if needed.

Search my calendar for the absence and return date, email threads for outstanding promises and requests, chats and other accessible messages for handover arrangements, and current project documents for the work people are likely to ask about. Search the relevant names and subjects across sources to establish what I owe, what has already been delivered, what has changed, and who has agreed to cover which responsibilities. Look at my sent messages to learn my normal level of warmth and directness. Use actual correspondence to establish a sender's name and relationship; do not invent familiarity.

Bring me the short picture you can establish: absence dates and time zone, likely incoming requests, confirmed commitments, and coverage. Distinguish a calendar entry from a confirmed vacation and a proposed deadline from an agreed promise. Ask focused questions, one at a time, to settle the dates, missing access, or consequential ambiguity. If I already supplied a fact, use it. If records cannot be accessed, investigate the supported connection first and request only the particular threads, handover note, or details needed to proceed.

Have me confirm the absence facts and the boundaries of the replies: who should receive them, what work information can be shared, which confirmed alternate contacts to mention, and which kinds of commitment I permit. Derive sensible proposed rules from the records and show examples so I can decide without designing the system myself. Do not assign coverage to a colleague just because their job title seems suitable.

Design the reply around what this person needs

For an eligible incoming message, read the message and relevant thread, identify the sender's actual request, and retrieve the smallest relevant set of prior correspondence and current project facts. Check later messages for changed commitments. Use that material to choose between providing an appropriate answer, referring to confirmed coverage, acknowledging an outstanding item, or giving a simple personal absence reply when little else is known.

Here is an illustration of the desired experience:

“Hi Kathy, I know I owe you that X report. I'm on vacation this week (this is being written by my AI agent) but I'll be back next Thursday and should be able to get it to you by the following Monday.”

Kathy, the report, and those dates are illustrative. Use a sentence like this only when the record and my confirmed directions support each claim. Being back on Thursday does not establish that I can deliver on Monday. If the delivery date is unknown, acknowledge the report and the confirmed return date without adding a new promise.

Keep replies short, warm, and useful. Personalization should reflect something relevant to the sender, not demonstrate how much of my correspondence you can retrieve. Reveal only information appropriate for this recipient. Include explicit AI disclosure in every automatic reply, with wording I approve, such as: “This reply was written by my AI assistant while I'm away. Please disregard anything that seems mistaken; I haven't personally reviewed it.” That disclosure does not replace checking the facts. When sources conflict or a request requires my judgment, give the approved absence information and acknowledge that I need to respond personally. Do not invent an answer or quietly make a decision on my behalf.

Implement a genuine automatic responder

Research the current official documentation for my provider and existing automation platform. Choose a route that can detect incoming mail, retrieve the permitted context, generate and check a reply, and send it from the agreed account without a manual step for every email. Inspect the real configuration and test its capabilities. Reading messages in our chat does not establish that a background service can read or send them.

Native vacation settings are useful to inspect, but their documented fixed message fields do not by themselves produce a different contextual answer for each sender. For Gmail, begin with vacation settings, incoming mailbox notifications, and sending messages. For Microsoft 365, begin with automatic reply settings, Outlook change notifications, and the message reply API. These were checked during authoring on September 13, 2026; verify the live documentation, account type, licensing, and tenant permissions when executing.

Build through an existing supported automation service when suitable; otherwise implement the necessary background service. For notification-based routes, handle subscription renewal, retrieval of actual message content, missed events, and repeated notifications according to the provider's documentation. A notification alone is not a complete incoming email. Preserve the correct account, recipient, and thread when replying; default to the sender rather than replying to everyone. Obtain the specific consent or administrator action the selected route requires, explain it in plain language, and do the remaining configuration yourself.

Keep sending disabled while building. Implement the agreed start and stop times, sender eligibility, reply frequency, and exclusions before any live activation. Avoid responses to our own messages, automated replies, bounces, mailing lists, spam, and other excluded traffic using available provider indicators and mail headers. Use persistent processing records and concurrency control to prevent repeated events from producing duplicate replies. Check whether the native vacation responder or another rule would also answer and prepare a coordinated switch, rather than leaving two responders active for the same messages.

Configure the workflow to record what it answered, the source facts used, and any failures, without exposing private context to recipients. Provide an easy pause control. When context retrieval or generation fails, use the approved simple absence response only if that fallback is part of the agreed rules; otherwise hold that reply and record the failure. The system as a whole must still be automatic. A collection of drafts is a rehearsal result, not completion of this activity.

Run a few meaningful tests and show me the results

Use a small set of representative historical messages without sending replies to their original senders, then use an explicitly agreed test address for actual delivery tests. Include a familiar sender with an outstanding commitment, a request that has a useful documented answer or alternate contact, and an unfamiliar sender for whom only a simple personal acknowledgment is supported. Also test an unsupported deadline, an automatic reply or duplicate event, and the end of the absence window. Choose compact cases that exercise several behaviors together.

Have a fresh reviewer, where available, inspect the exact incoming messages, source excerpts, agreed rules, and generated replies without your defense of them. Ask them to find invented promises, wrong dates, inappropriate disclosures, misleading familiarity, and failures to address the request. Give bounded case packets and have one coordinator reconcile findings and control all sends. Without subagents, review in a separate pass and disclose that it shares context.

Check both wording and operation: source-supported claims, accurate recipient and thread, AI disclosure, one actual delivered reply for an eligible test, and no reply for an excluded case. Verify the start/stop behavior and the pause control. Repair material failures, rerun affected tests, and show me a few complete before-and-after examples with a concise statement of what passed and what remains unresolved.

Ask, activate, and verify

When the tests support activation, show the exact sending scope, schedule, approved disclosure and pause control. Then ask exactly: “Are you confident enough to turn this on?” Wait for my answer. Do not turn it on while waiting. If I say no or ask for changes, revise and retest those changes; leave automatic sending off until I explicitly approve activation.

After I say yes, activate the agreed workflow and coordinate the existing vacation settings as approved. Verify the saved configuration and an actual incoming-to-delivered reply using the agreed test sender. If the absence begins later, verify the armed schedule and a controlled test of the same sending path, then restore the intended schedule and report that it is armed for those dates. Do not claim a future event has already happened.

Give me a concise record of what is active or scheduled, the absence dates, what kinds of reply it sends, where I can inspect replies, how to pause it, and how it will stop or restore the prior setup. A “send accepted” response alone does not prove inbox delivery. If permissions or deployment prevent activation, identify the exact blocker and preserve the tested work as unfinished. The completed result is a tested personalized automatic out-of-office responder, activated or verifiably armed for the confirmed absence after my yes.

Done when

After you have reviewed the tests and said yes, the personalized responder is active or verifiably scheduled for your confirmed absence. Actual delivery has been tested, every automatic reply discloses AI authorship, and you know how to inspect replies and pause the system. Unavailable activation access means the setup remains unfinished.

These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.