how to

How to get AI to write email in your style (without re-prompting it every time)

the short answer

You can train AI on your writing style two ways: manually, by pasting samples and describing your tone in a prompt every time, or automatically, by letting a tool learn from your existing writing; echo takes the automatic route — it studies your sent Gmail to model your tone, phrasing, and sign-offs, then sharpens with a one-tap good-draft / needs-work signal, so you don't re-prompt it for every email.

If you've tried to make AI write like you, you know the manual version: collect a few of your emails, paste them into a chat, and add an instruction like 'match this tone'. It works for one message, then the context is gone and you do it all again on the next one. It's training, but it doesn't stick.

There's a better way to think about style learning, and a lower-effort way to do it. This guide covers what 'writing style' even means to a model, the two ways to teach it, and how echo handles it automatically by learning from the email you've already sent.

no promptingecho learns your style from sent mail instead of asking you to describe it
open echo →about echo

What 'writing style' is to a model

To a model, your style is a set of statistical tendencies it can imitate: how formal you are, how long your sentences and replies run, the words and phrasings you favour, how you open and close. It can't read your mind, but given enough of your real writing it can approximate those patterns surprisingly well.

The key word is real writing. Describing your style in adjectives ('friendly, concise') is a weak signal; showing the model dozens of things you've actually written is a strong one. The more authentic examples it has, the closer the imitation.

Two ways to teach it

The manual route is sampling and prompting: gather examples, paste them in, describe the tone, and re-do that context for every new message. It's flexible and free, but it's fiddly and it forgets, so the quality depends on you redoing the work each time.

The automatic route is learning from a corpus you already have. Instead of you assembling samples, the tool reads your existing writing and builds a lasting profile. echo uses your sent mail for this, so the 'training set' is just the email you've already sent — no collecting, no pasting.

How to keep the style sharp

However it's trained, a style model gets better with feedback. The fastest signal is telling it when a draft nailed your voice and when it missed, so it can adjust toward the former.

echo builds that in with a one-tap good-draft / needs-work rating on each draft. You don't manage settings or rewrite prompts — you just react to drafts as they come, and the model tunes itself toward how you actually sound.

how it works

  1. 01

    Connect your Gmail

    echo reads your sent mail — the authentic writing sample that teaches it your style.

  2. 02

    Let it build a profile

    It models your tone, phrasing, length, and sign-offs from your real replies, no pasting required.

  3. 03

    Generate drafts

    echo writes replies in that learned style for emails that need a response.

  4. 04

    Rate good vs needs-work

    Your one-tap feedback nudges future drafts closer to your voice.

  5. 05

    Edit and send

    Approve each reply yourself — the human check stays, and the style keeps improving.

frequently asked

Do I need to collect writing samples to train echo?
No. echo learns from your sent Gmail automatically, so your existing emails are the training data — there's nothing to gather or paste.
How much email does it need to learn my style?
It learns from your sent mail, so the more you've written, the richer the profile. The good/needs-work signal then refines it from there.
Is training AI on my style the same as fine-tuning a model?
Not in the heavy technical sense. echo builds a voice profile from your writing and uses feedback to refine drafts — you don't manage any model training yourself.
Can I correct it if a draft sounds wrong?
Yes — edit the draft and tap needs work. That feedback teaches echo what to avoid and pulls future drafts toward how you actually write.

Last updated June 9, 2026

more on echo

ready to try echo?

open echo →