Five complete templates for emails, documents, forecasts, marketing plans and advice.
To copy and paste into your AI
Ready-made prompts for everyday work
Five complete templates, built with the six steps. Copy, paste into your AI and replace the parts in square brackets. The more context you add, the better the draft.
Writing an email
Act as my assistant in writing professional emails.
Context: I am [role] at [company, industry, country]. I am writing to [recipient: role and relationship to me, for example client, colleague, supplier, manager]. The situation is this: [what happened, in 2–3 sentences]. My relationship with this person is [new / established / strained].
Goal: write a draft email that achieves [specific result: a reply by a date, a meeting, a yes, clearing up a misunderstanding].
Requirements:
1. In the first line, the reason for the email.
2. Include these points: [points to cover].
3. End with a clear request and, if needed, a date.
Limits:
- Do not invent facts, dates, figures or commitments I have not given you: where they are missing, write [to be completed].
- No stock phrases ("I hope this email finds you well") and no overly enthusiastic tone.
- Do not promise anything on my behalf beyond what I have written to you.
Format: subject line + text, at most [number] lines, tone [formal / friendly / direct], language [English / German / French / Italian / ...], [first-name basis / formal address].
Before you write, tell me if you are missing anything to hit the goal. Then suggest two versions: one more direct and one more diplomatic.
Writing a document
Act as my assistant in drafting work documents.
Context: I am [role] at [company, industry, country]. The document is for [purpose] and will be read by [recipient, and what they already know about the subject]. Starting material: [paste notes, data, texts, or write "none"].
Goal: prepare a draft [type: report, procedure, proposal, minutes, fact sheet] on [subject].
Requirements:
1. Propose the table of contents first and wait for my confirmation before writing the rest.
2. Use only the information I have given you and keep facts separate from your proposals.
3. At the end, list the decisions to be made and who needs to make them.
Limits:
- Do not invent data, sources, names or quotations: where they are needed, write [to be verified].
- No generic sentences: every paragraph must say something specific about our case.
- If a part requires technical expertise (legal, tax, security, medical), flag it instead of improvising.
Format: [length], short headings, short paragraphs, tone [tone], language [language].
Before you start, ask me the questions you need.
Making a forecast
Act as my assistant in analysing data and scenarios.
Context: I am [role] at [company, industry, market]. I want to estimate [what: sales, costs, cash flow, demand] for [period]. Data I have: [paste the historical data with units and period, or attach the file]. Things I already know will change: [new products, prices, seasonality, events].
Goal: prepare an estimate with three scenarios: conservative, likely, optimistic.
Requirements:
1. For each scenario, state the assumptions clearly: what goes up or down, by how much and why.
2. Show the calculations step by step, so that I can redo them in a spreadsheet.
3. Name the 3 factors that change the result the most and what to check every month.
Limits:
- Use only the data I have given you. Do not invent industry averages, growth rates or statistics: if you need them, ask me or write [data to find, and where to look for it].
- Do not present the estimate as certain: tell me how reliable it is and why.
- If there is too little data for a serious estimate, tell me before doing the calculations.
Format: one table per scenario and 5 lines of comment, understandable for someone who is not an analyst.
Preparing a marketing plan
Act as my assistant in marketing planning.
Context: [company], industry [industry], market [country or region, B2B or B2C]. We sell [product or service] to [typical customer: who they are, what they need, how they decide]. Main competitors: [names]. Budget: [amount and period]. Internal resources: [people, skills, tools]. What we have already done and how it went: [actions and results].
Goal: prepare a draft marketing plan for [period] that leads to [measurable goal: leads, sales, awareness among a specific audience].
Requirements:
1. Start from the goal and the typical customer, not from the channels.
2. For each action: channel, message, estimated cost, who does it, indicator to measure it.
3. A month-by-month calendar, and what to change if results do not come after [number] weeks.
Limits:
- Do not propose actions that do not fit the budget or resources I have given you.
- Do not invent benchmarks, average costs or conversion rates: if you use them, write [estimate to be verified].
- If you propose actions involving personal data (newsletters, contact lists, profiling), flag what needs to be checked for privacy.
Format: one table and a half-page summary for management.
Before you start, ask me 3 questions about the typical customer.
Asking for advice
Help me think through a decision. Do not decide for me.
Context: I have to decide [decision]. The situation is this: [facts, constraints, timing]. The options I see: [A, B, ...]. What matters most to me is [criteria: cost, time, risk, relationships, values].
Goal: help me compare the options.
Requirements:
1. For each option: advantages, risks and what happens if it goes wrong.
2. The information I am missing to decide well, and where to find it.
3. The questions an expert on the subject would ask me before giving an opinion.
Limits:
- Do not just agree with me: if my premise is wrong, tell me.
- Distinguish what you know from what you assume.
- If the decision concerns health, law, taxes, investments or safety, tell me clearly which professional I should consult.
Format: a comparison table and a 5-line summary. No final recommendation until I have answered your questions.
If it is not your field, ask the expert.Because people who do not know the job do not know what to ask, or what is missing from the answer: that is what we saw tonight, case after case. The AI writes well even when it is wrong, and only someone who knows the subject will notice.