Start by Defining Your Style
Before you can teach an AI, you need to understand your own writing style. Are you warm and concise, or formal and detailed? Take a few minutes to articulate your professional tone. Write down key characteristics. For instance, your personal style guide
might include notes like: "Professional but friendly, avoids jargon, uses short paragraphs, and always opens with 'Hi [Name]'." You can also list things to avoid, such as overly casual slang or robotic phrases like "per my last email." This foundational document becomes the rulebook you will give the AI.
Curate Your Best Examples
AI learns best from clear examples. Scour your sent folder and gather a handful of emails that perfectly represent the tone you want to achieve. Find 3-5 'gold standard' examples: a project update to your team, a follow-up with a client, and a networking request. These samples provide the AI with concrete evidence of your style in different contexts. Instead of just telling the AI to be 'professional,' you are showing it what your version of professional looks like in practice. This is far more effective than abstract descriptions alone.
Use Custom Instructions and Prompts
Most modern AI writing tools, including browser extensions, have a 'Custom Instructions' or 'Style Guide' feature. This is where you combine your defined style and examples. In the instruction field, paste your style guide notes. Then, you can use a prompt that directs the AI to use these rules. A good starting prompt is: "You are my email writing assistant. Using my style guide, rewrite the following draft to sound like me. Ensure the tone is [your keywords], the sentences are [your preference], and the closing is always [your sign-off]." This command-and-control approach sets clear boundaries for every task.
Teach What to Avoid
Just as important as showing the AI what to do is showing it what not to do. AI models are often trained on a vast and varied dataset, which includes a lot of overly formal or clichéd business-speak. If you find the AI using phrases you dislike, such as "in today's fast-paced world" or "I hope this email finds you well," explicitly tell it to stop. You can add a section to your custom instructions called "Phrases to Never Use." Providing this negative feedback helps the model narrow its options and avoid common AI writing pitfalls, making the output sound less generic and more like a real person.
Master Context-Specific Prompts
A single prompt rarely works for every situation. To get the best results, you need to provide context for each specific email. Before pasting your draft for a rewrite, tell the AI about the audience and your goal. For example: "Rewrite this draft for a new client I'm trying to build a relationship with. The tone should be extra welcoming and clear." Or, "Shorten this draft for an internal update to my busy manager. Keep it direct and under 100 words." Giving the AI context about the recipient and desired outcome helps it make smarter decisions about tone, length, and formality.
Iterate, Edit, and Refine
Training an AI is not a one-time setup; it's an ongoing process. The first few drafts the AI produces will likely need human editing. Don't just fix the text in your email client—use your edits as a teaching moment. If the AI's draft is close but not quite right, provide feedback in your next prompt. For example: "That was good, but let's try again. Make it less formal and remove the bullet points." This iterative feedback loop is crucial. The more you guide and correct the AI, the better it will become at predicting your needs and matching your authentic voice from the very first draft.
