The three emails you keep putting off, each under 120 words

There is a kind of email that sits in your drafts folder for a week. Not because it is long, but because getting the tone wrong costs you something: a manager who now thinks you are difficult, a landlord who stops replying, a neighbor you have to pass in the hallway every morning.

This guide covers three of those emails. Each one has a prompt you can paste into any free chatbot, a walkthrough of what to put in the square brackets, and a description of what the reply should look like so you know when to send it and when to ask again.

The prompts come from the chapter on awkward correspondence in the book Stop Typing, Start Asking. They are reproduced exactly. Everything around them is about the situation, not the software.

Why 'write a polite email' gets you nothing usable

The naive version of this request is one line: write a polite email telling my boss I can't take on the audit report. The chatbot will oblige, and what comes back is three paragraphs of apology, a vague promise to circle back, and no actual decision for anyone to make. It reads polite. It does not do the job.

The problem is not the model. The one-line request leaves out the four things that make this email hard: who is asking, what specifically they want, what it would displace, and where your line is. Without those, the chatbot fills the gaps with softening language, because softening is what polite looks like when there are no facts to be polite about.

The three prompts below all fix this the same way. They list the facts as labelled fields, they state the tone in plain adjectives, they ban the words that make you sound weak, and they cap the length. The cap matters more than it looks: a 90-word note gets read; a 300-word one gets skimmed and resented.

Saying no to extra work without apologising

The situation: a manager or colleague sends you something that will not fit. You want to say no, but you also want to keep the relationship and not look like someone who guards their time at the team's expense.

The trick in this prompt is the third constraint. It tells the chatbot to frame your reply as a trade-off for the sender to decide, not as you refusing. That turns 'I can't' into 'here is what it costs, you choose', which is a sentence a manager can actually act on.

Draft a concise, professional email responding to a work request that exceeds my current capacity.
Context:
- Sender: [Manager's Name / Colleague's Name]
- Incoming Request: [Briefly describe the requested task, e.g., Take over the quarterly client audit report]
- Current Priority: [Describe your existing primary deliverables, e.g., Completing the Q3 software migration rollout by Friday]
- Boundary: [Cannot take this on without delaying primary deliverables / Do not monitor emails after 6:00 PM]

Constraints:
- Tone: Professional, collaborative, calm, and firm.
- No unnecessary apologies (do not use "I'm sorry" or "Unfortunately").
- Frame the situation as a priority trade-off decision for the sender rather than personal unwillingness.
- Keep the length under 100 words.

Fill the four context lines with real specifics. Sender is a name. Incoming Request is the task in a few words, exactly as they phrased it. Current Priority is the thing you are already committed to, with its deadline, because the deadline is what makes the trade-off concrete. Boundary is the line you will not cross; the book gives two examples, one about capacity and one about hours, and you should pick the one that matches, not both.

A good reply is under 100 words, names the existing deadline, quantifies the delay the new task would cause, and ends by handing the decision back. The book's example does this in five sentences and never uses the word sorry. If the draft you get opens with 'Unfortunately' or offers to squeeze it in, the chatbot ignored the constraints. Paste the prompt again and add one line at the end: follow every constraint exactly.

Getting a landlord to fix the thing they said they would fix

The situation: you reported a leak, a broken heater, a lock that does not lock. Someone said they would handle it. Nothing happened. Now you need a second message that is firmer than the first without becoming a threat.

What makes this prompt different from the other two is that it is building a paper trail. The constraint about establishing a written record means the email states, in writing, that you already reported this and when. If the issue ever goes further, that sentence is the one that matters. The request for written confirmation of an appointment does the same job from the other side: it forces a reply that either commits to a date or visibly avoids doing so.

Draft a formal, polite maintenance escalation email to a property manager or landlord.
Details:
- Recipient: [Landlord / Property Management Company Name]
- Property Address: [Unit Number and Street Address]
- Maintenance Issue: [Describe issue, e.g., Water leaking under the kitchen sink]
- First Reported Date: [Date of initial report, e.g., October 12th via text]
- Impact: [Describe functional or safety impact, e.g., Unable to use kitchen sink; risk of cabinet wood rot]
- Required Action: [Request scheduled repair technician within 48 hours]

Constraints:
- Tone: Civil, objective, fact-focused, and legally clear.
- Explicitly establish a written record of previous notifications.
- Request written confirmation of the repair appointment date and time.
- Keep under 120 words.

Be exact with dates and channels in First Reported Date. 'October 12th via text' is better than 'a couple of weeks ago', because the whole point is the record. In Impact, describe what you cannot do and what is getting worse, in neutral words. 'Unable to use kitchen sink; risk of cabinet wood rot' is the right register. 'This is unacceptable' is not, and the tone constraint will push against it anyway. Required Action should be a specific window: 48 hours, or a named day.

The reply you want is under 120 words, opens by restating the original report date, describes the current state and the risk in one or two sentences, and asks for written confirmation by a deadline. Check the draft for two things before sending: that the property address is complete, and that it does not accuse anyone of anything. Fact, impact, request. Nothing else.

The noise note that does not start a feud

The situation: someone next door is loud at the wrong hours. You have decided that knocking on the door is worse than a note, but a note can go badly too. Too stiff and it reads like a complaint to management; too casual and it gets ignored.

The constraint to assume good intentions is doing most of the work here. It makes the note open with the physical fact that sound carries between units, which gives the neighbor a way to fix things without having been in the wrong. That is what non-accusatory means in practice.

Draft a polite, friendly, but direct written note to a neighbor regarding recurring noise.
Details:
- Specific Issue: [Describe noise, e.g., Loud television bass / dog barking late at night]
- Timing: [Specify hours, e.g., Between 11:00 PM and 2:00 AM on weeknights]
- Impact: [Brief neutral impact, e.g., Wakes up family members / disrupts sleep before workdays]
- Goal: [Request volume reduction during designated quiet hours]

Constraints:
- Tone: Neighborly, empathetic, calm, and non-accusatory.
- Assume good intentions (assume they are likely unaware of sound transfer between units).
- Provide a polite closing that leaves room for open communication.
- Length: 60 to 90 words.

Keep Specific Issue to the sound itself, not a theory about the person. 'Loud television bass' is fine; 'your parties' is not. Timing should be the actual hours you have noticed, because a precise window is easier to agree to than 'late at night'. Impact is one neutral clause. Goal is what you are asking for, and it helps to ask for something small and specific: volume down after a certain time, or headphones on weeknights.

Expect 60 to 90 words: a warm opener, the sound-carries observation, the specific hours, one direct request, and a closing that invites them to talk to you. Read it once as if you were the neighbor. If any sentence would make you defensive, delete it and regenerate. Sign it with your unit number so they know where it came from, and leave out anything you would not want read aloud in the lobby.

Before you send any of them

All three drafts need the same two-minute edit. Replace every bracket the chatbot left in. It will often keep a placeholder like the manager's name or your phone number literally, and sending that is worse than sending nothing.

Then check the length against the cap in the prompt. Chatbots drift long. If the reply is over, ask it to cut to the limit rather than trimming by hand, because it will usually remove the padding and keep the facts.

Finally, read the subject line. The book's examples are specific: priority alignment regarding the audit report, maintenance escalation for an unresolved leak. A subject that names the thing gets opened and gets filed. 'Quick question' does neither.

What this cannot do

• It cannot know your local tenancy rules. The landlord email creates a record and asks for a date. It does not tell you what your jurisdiction requires a landlord to do, in what timeframe, or what your options are if they do not. Look that up separately before you rely on any deadline in the draft.
• It cannot read the relationship. The prompts assume a manager who responds to trade-offs and a neighbor who responds to courtesy. If you already know that is not true in your case, the tone constraints are a starting point to adjust, not a guarantee.
• It cannot decide the boundary for you. The work email only works if you have actually decided what you will not do. Paste a vague boundary and you get a vague email back.
• It will not follow every constraint every time. Word caps and banned words are the instructions chatbots ignore most often. Count the words and search for the banned ones before you send.
• It cannot send the email or follow up. An escalation only works if there is a next step. Decide now what you will do if there is no reply in 48 hours, because the draft will not do that for you.

Related reading

• Stop Typing, Start Asking — full first chapter, free, no email required
• Full contents and what is included
• The one-page AI prompt checklist
• How to repurpose one long-form video into 20 posts
• All Avalon Store products