Study sent emails to build your style guide.
Your sent mail contains a style guide you have never written. The interesting part is how recognizably you remain yourself while writing differently to an old collaborator, a new customer, or someone you need to disagree with. Have AI study those differences, build a guide you recognize, and investigate how your email setup could use it to prepare drafts.
Try it now
Copy the prompt into your AI tool of choice, then answer its questions as they come.
Paste it into any AI chat to start. For most people, this is all you need.
Everything on this page in one paste: the prompt plus the full plan, every step and what to check. Use it for longer moves, to plan ahead, to share with a colleague, or if your AI can't open links.
How it goes
What the AI will do
- Check access to your sent email and relevant chats, then study your own writing across recipients, situations, and time.
- Build and test a reusable voice guide, then investigate how your actual email and AI setup could use it to generate drafts.
What to have ready
- Access to sent messages where available, or a few interesting emails pasted or exported to begin; a larger, more varied collection supports a stronger guide.
- Context about recipient relationships when the records do not explain them, and a couple of messages you need to write.
Where you come in
- Say which observed habits feel like you and which you would like to change; correct sample drafts for two different recipients.
- Review the integration proposal and decide whether you want help implementing it. You decide when an email is ready to send.
What you'll get
A reusable voice guide with recipient-specific variations, corrected example drafts, and a concrete account of how to use it in your email workflow, including what has actually been tested.
You're done when
You have reviewed a reusable voice guide and corrected example drafts for two recipient relationships. You also have a concrete way to use the guide for email drafting, with proposed, tested, and unavailable integration capabilities clearly distinguished. A small source sample produces an explicitly provisional guide.
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. Help me build a practical writing style guide from my sent emails and use it to prepare emails that sound like me. Pay particular attention to how my voice changes with the recipient. Carry this through a tested guide and an investigation of how my actual setup could generate draft emails with it. ### 1. Start with access to my writing First inspect and try available access to email and relevant messages, including chats. Say which sources you can actually read and whether you can retrieve sent messages and their threads. This activity works best with a substantial collection of my own sent writing. If you cannot access it, say so, investigate current connection or export options for my environment, then ask whether there is a way to enable access. Preserve any permission and source boundaries already given. If the company lacks a suitable connection, explain what read-only access would make possible and offer to draft an IT request. If access cannot be enabled, help me paste or export a few interesting sent emails. Welcome a small beginning; explain which additional kinds of examples would strengthen it rather than demanding a prepared packet. Retain recipient context, dates, and the surrounding exchange where available. An interview can explain intent, but cannot substitute for examples of my writing. With only a few examples, produce a provisional guide and name the relationships or situations it has not yet been tested against. ### 2. Build a varied, attributable sample Use available sent folders and search tools to find writing across recipients, relationships, purposes, channels, and a useful recent period. Include, where present, routine requests, explanations, encouragement, persuasion, disagreement, apologies, decisions, and follow-ups. Do not let one long thread or frequent recipient dominate the sample. Add older examples where useful to see whether a habit still represents me. Read enough of each thread to understand what I was responding to and what I wanted. Consult contacts, roles, prior correspondence, and relevant meeting or project records to establish who recipients are and our relationship. Ask only about consequential gaps; an address or job title alone does not establish personal closeness. Separate my newly composed text from quotations, forwarded text, signatures, templates, automated messages, and other people's writing. Flag uncertain authorship, including material that may have been drafted for me, before using it as evidence. Keep a compact coverage record with dates, recipients or relationship categories, purpose, source references, and gaps. If the collection exceeds your capacity, assign bounded recipient or purpose groups to agents with complete exchanges, the analysis questions below, and instructions to return sourced observations, excerpts, counterexamples, and coverage. Reconcile these yourself. Without delegation, use successive bounded passes and the same coverage record. ### 3. Work out what makes the voice mine Examine actual choices, with short excerpts that show them: - Vocabulary and register; sentence length, rhythm, and variation. - Paragraphing, overall structure, and how quickly I reach the point. - Openings and endings; warmth, gratitude, encouragement, and humor. - Directness, hedging, confidence, and expressions of uncertainty. - Questions and invitations; how I ask for work or make commitments. - Disagreement, correction, and gentle redirection. - Examples, evidence, punctuation, and formatting. These are lenses, not a scoring system or traits you must find. Compare similar purposes across different recipients and different purposes with the same recipient. Distinguish enduring choices from a topic's vocabulary, urgency, channel conventions, and the demands of a particular relationship. Look for recurring patterns and contrary examples. A memorable phrase in one email is not automatically my defining habit. Present a concise account of my core voice and its contextual variations, grounded in passages I can inspect. Separate well-supported observations from tentative ones. Ask which habits I want to preserve or change, and wait for my answer before making the guide prescriptive. Keep observed habits and my desired changes distinct. ### 4. Write and test the reusable guide Create a standalone guide an AI could use without this conversation: the core voice, adjustments by evidenced recipient relationship and purpose, concrete writing directions, and short examples showing the directions in action. Explain what context a drafting AI needs: recipient, relationship, thread, purpose, facts, requested action, and any commitments I intend. Tell it to ask when essential information is missing and never invent personal experiences, facts, or promises to make a draft sound natural. Use two actual writing needs for different recipient relationships, if available. Search their relevant exchanges and project context before asking me for missing intent or facts. Otherwise ask me for two situations, or use clearly labeled practice situations that I agree are representative. Produce full sample emails from the guide and known facts. Ask what sounds unlike me or mishandles the relationship. Revise the guide and drafts from that feedback, then check both examples again. Save the useful corrected examples with the guide; do not present an unreviewed imitation as my accepted voice. ### 5. Investigate integration with my actual email workflow Inspect the email application, AI environment, and usable connections I actually have. Use current official documentation to establish whether and how that setup can retrieve thread context, retain or retrieve the guide, generate an email, and place it in a draft for review. Check the relevant capabilities where available; label documented but untested capabilities accurately. Ask for only the setup details you cannot inspect. Explain a concrete route in plain language: what starts a draft, where the guide lives, how the recipient and thread enter the task, where the resulting draft appears, what I review, and what setup or permissions are required. Distinguish an existing verified integration, a feasible proposed setup, and manual copy/paste. If direct integration is unavailable, give me a self-contained drafting prompt containing the accepted guide and directions to supply the thread, purpose, and facts; make the fallback immediately usable. If I request implementation, complete the supported setup you can perform, guide me through any necessary action only I can take, and test one real draft. Inspect whether the draft uses the right thread and voice guide and appears in the promised location. Report what passed and what remains unverified. Sending stays my decision; do not send as part of this activity. Deliver the final guide, corrected examples, reusable drafting instructions, and the integration explanation together. If implementation was not requested, finish with the usable guide and concrete setup proposal; do not describe the proposal as installed. 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.
View full AI instructions
Study sent emails to build your style guide.
What the AI will do
- Check access to your sent email and relevant chats, then study your own writing across recipients, situations, and time.
- Build and test a reusable voice guide, then investigate how your actual email and AI setup could use it to generate drafts.
What you can provide
- Access to sent messages where available, or a few interesting emails pasted or exported to begin; a larger, more varied collection supports a stronger guide.
- Context about recipient relationships when the records do not explain them, and a couple of messages you need to write.
Where you'll take part
- Say which observed habits feel like you and which you would like to change; correct sample drafts for two different recipients.
- Review the integration proposal and decide whether you want help implementing it. You decide when an email is ready to send.
What you'll get
A reusable voice guide with recipient-specific variations, corrected example drafts, and a concrete account of how to use it in your email workflow, including what has actually been tested.
Activity context
Your sent mail contains a style guide you have never written. The interesting part is how recognizably you remain yourself while writing differently to an old collaborator, a new customer, or someone you need to disagree with. Have AI study those differences, build a guide you recognize, and investigate how your email setup could use it to prepare drafts.
For the user
Use this in the AI environment where you work with email. It will first check what it can read and help with another route if needed. You will recognize or correct its account of your voice, test it on real writing, and review how the guide could fit into drafting emails.
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
Help me build a practical writing style guide from my sent emails and use it to prepare emails that sound like me. Pay particular attention to how my voice changes with the recipient. Carry this through a tested guide and an investigation of how my actual setup could generate draft emails with it.
1. Start with access to my writing
First inspect and try available access to email and relevant messages, including chats. Say which sources you can actually read and whether you can retrieve sent messages and their threads. This activity works best with a substantial collection of my own sent writing. If you cannot access it, say so, investigate current connection or export options for my environment, then ask whether there is a way to enable access. Preserve any permission and source boundaries already given. If the company lacks a suitable connection, explain what read-only access would make possible and offer to draft an IT request.
If access cannot be enabled, help me paste or export a few interesting sent emails. Welcome a small beginning; explain which additional kinds of examples would strengthen it rather than demanding a prepared packet. Retain recipient context, dates, and the surrounding exchange where available. An interview can explain intent, but cannot substitute for examples of my writing. With only a few examples, produce a provisional guide and name the relationships or situations it has not yet been tested against.
2. Build a varied, attributable sample
Use available sent folders and search tools to find writing across recipients, relationships, purposes, channels, and a useful recent period. Include, where present, routine requests, explanations, encouragement, persuasion, disagreement, apologies, decisions, and follow-ups. Do not let one long thread or frequent recipient dominate the sample. Add older examples where useful to see whether a habit still represents me.
Read enough of each thread to understand what I was responding to and what I wanted. Consult contacts, roles, prior correspondence, and relevant meeting or project records to establish who recipients are and our relationship. Ask only about consequential gaps; an address or job title alone does not establish personal closeness.
Separate my newly composed text from quotations, forwarded text, signatures, templates, automated messages, and other people's writing. Flag uncertain authorship, including material that may have been drafted for me, before using it as evidence. Keep a compact coverage record with dates, recipients or relationship categories, purpose, source references, and gaps.
If the collection exceeds your capacity, assign bounded recipient or purpose groups to agents with complete exchanges, the analysis questions below, and instructions to return sourced observations, excerpts, counterexamples, and coverage. Reconcile these yourself. Without delegation, use successive bounded passes and the same coverage record.
3. Work out what makes the voice mine
Examine actual choices, with short excerpts that show them:
- Vocabulary and register; sentence length, rhythm, and variation.
- Paragraphing, overall structure, and how quickly I reach the point.
- Openings and endings; warmth, gratitude, encouragement, and humor.
- Directness, hedging, confidence, and expressions of uncertainty.
- Questions and invitations; how I ask for work or make commitments.
- Disagreement, correction, and gentle redirection.
- Examples, evidence, punctuation, and formatting.
These are lenses, not a scoring system or traits you must find. Compare similar purposes across different recipients and different purposes with the same recipient. Distinguish enduring choices from a topic's vocabulary, urgency, channel conventions, and the demands of a particular relationship. Look for recurring patterns and contrary examples. A memorable phrase in one email is not automatically my defining habit.
Present a concise account of my core voice and its contextual variations, grounded in passages I can inspect. Separate well-supported observations from tentative ones. Ask which habits I want to preserve or change, and wait for my answer before making the guide prescriptive. Keep observed habits and my desired changes distinct.
4. Write and test the reusable guide
Create a standalone guide an AI could use without this conversation: the core voice, adjustments by evidenced recipient relationship and purpose, concrete writing directions, and short examples showing the directions in action. Explain what context a drafting AI needs: recipient, relationship, thread, purpose, facts, requested action, and any commitments I intend. Tell it to ask when essential information is missing and never invent personal experiences, facts, or promises to make a draft sound natural.
Use two actual writing needs for different recipient relationships, if available. Search their relevant exchanges and project context before asking me for missing intent or facts. Otherwise ask me for two situations, or use clearly labeled practice situations that I agree are representative. Produce full sample emails from the guide and known facts. Ask what sounds unlike me or mishandles the relationship. Revise the guide and drafts from that feedback, then check both examples again. Save the useful corrected examples with the guide; do not present an unreviewed imitation as my accepted voice.
5. Investigate integration with my actual email workflow
Inspect the email application, AI environment, and usable connections I actually have. Use current official documentation to establish whether and how that setup can retrieve thread context, retain or retrieve the guide, generate an email, and place it in a draft for review. Check the relevant capabilities where available; label documented but untested capabilities accurately. Ask for only the setup details you cannot inspect.
Explain a concrete route in plain language: what starts a draft, where the guide lives, how the recipient and thread enter the task, where the resulting draft appears, what I review, and what setup or permissions are required. Distinguish an existing verified integration, a feasible proposed setup, and manual copy/paste. If direct integration is unavailable, give me a self-contained drafting prompt containing the accepted guide and directions to supply the thread, purpose, and facts; make the fallback immediately usable.
If I request implementation, complete the supported setup you can perform, guide me through any necessary action only I can take, and test one real draft. Inspect whether the draft uses the right thread and voice guide and appears in the promised location. Report what passed and what remains unverified. Sending stays my decision; do not send as part of this activity.
Deliver the final guide, corrected examples, reusable drafting instructions, and the integration explanation together. If implementation was not requested, finish with the usable guide and concrete setup proposal; do not describe the proposal as installed.
Done when
You have reviewed a reusable voice guide and corrected example drafts for two recipient relationships. You also have a concrete way to use the guide for email drafting, with proposed, tested, and unavailable integration capabilities clearly distinguished. A small source sample produces an explicitly provisional guide.
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.
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.