Question 1

Won't it sound generic?

Accepted Answer

No — drafting follows the voice you set in your Personal Context and your context engine, so replies read like you wrote them, with real facts pulled in rather than invented.

Question 2

Which AI model writes my drafts?

Accepted Answer

You choose — a fast, balanced, or powerful model in Settings → AI & Voice — or bring your own key with BYOK and run on your own credits with no caps.

Question 3

How do reply templates work?

Accepted Answer

Save replies you send often, then insert them by typing // in the composer or from the toolbar. Placeholders like {{first_name}} resolve from your context automatically.

Question 4

How does AI write in my voice?

Accepted Answer

AI Emaily drafts in the voice you set in your Personal Context — your tone, phrasing and sign-offs — plus per-client profiles for how formal you are with a client versus a cofounder. When you hit "Draft in my voice" on a thread, it composes a reply in that tone and pulls the real facts from your context engine rather than inventing them. There's no model to train by hand; set your voice once, tune it per client, and each reply sounds like you actually wrote it.

Question 5

Will AI-written emails sound generic?

Accepted Answer

No — that's the core difference. Generic AI email is trained on everyone; AI Emaily's drafting follows the voice you set in your Personal Context and your context engine, so replies carry your phrasing and your real facts. Recipients can usually tell when something was bolted-on AI; the point here is that they can't. And nothing is final on autopilot of trust — you review every draft, reshape it inline, and only you (or bounded Autopilot within your rules) decide when it sends.

Question 6

Can I change the tone of a draft?

Accepted Answer

Yes, in one click. The composer gives you a row of inline AI tools that apply in place: rewrite, shorten, lengthen, make it more formal or more casual, or translate. So you can soften a reply to a client, tighten one to a cofounder, or flip a draft to another language without leaving the editor. It's editing at the speed of thought — you start from a draft that already sounds like you, then nudge it rather than facing a blank page.

Question 7

Does it learn from my sent emails?

Accepted Answer

It can, and you control it. Your voice starts with what you set in your Personal Context — your tone, phrasing and sign-offs — plus per-client profiles for how it shifts per person. Style Memory, on by default, also builds a private style card (greetings, sign-offs, rhythm, formality) from up to 40 of your own sent emails, used only for your drafts; turn it off any time in Settings → AI & Voice → Match my writing style. We never train models on your email, and AI calls run with zero-retention from providers so your content isn't stored or reused.

Question 8

Can it write emails in other languages?

Accepted Answer

Yes. The inline translate tool converts any draft to another language in a click, right in the composer, so you can reply to international contacts in their language without leaving your inbox. Drafting and the wider app also ship in 21 languages. Translation runs through the same grounded, zero-retention AI pipeline as the rest of drafting, so your content isn't stored or used to train any model in the process.

Question 9

How do I draft a reply from scratch fast?

Accepted Answer

Two ways. On any thread, "Draft in my voice" turns the conversation into a complete reply in your tone with real names, numbers, and dates filled from your context — no blank page. Or jot a few fragments and let idea-to-email expand them into a full, on-voice message, pulling relevant detail from the thread and your inbox. Then reshape it inline — shorten, re-tone, translate — and send. Most replies go from nothing to ready in seconds.

Question 10

Can I use my own AI model or API key?

Accepted Answer

Yes. In Settings → AI & Voice you pick the model behind drafting — fast, balanced, or powerful — to trade speed against quality. Prefer your own account? BYOK lets you bring your own provider key and run drafting on your own credits with no metered caps. Your key is treated as a crown jewel: envelope-encrypted, decrypted only inside an isolated worker, never exposed client-side or written to logs. All AI runs zero-retention, so providers don't keep your content.

Question 11

Are my drafts private and not used for training?

Accepted Answer

Yes. We never train models on your mail or your drafts, full stop. AI calls go through the gateway with zero-retention from model providers, so prompts and content aren't stored or used to improve anyone's model. Your voice settings are used only to write your own replies. If you want your data to never leave your provider account at all, BYOK runs drafting on your key, decrypted only in an isolated worker and never logged.

Question 12

How accurate are the facts in AI drafts?

Accepted Answer

As accurate as your context, because the agent doesn't guess. Names, numbers, links, dates, and account status come from your context engine — client profiles and typed variables you define once — so drafts use real values rather than plausible inventions. When a fact isn't in context, it won't fabricate one. A 'did-you-forget' check also flags commitments and questions you left unaddressed before you send, so the reply is not just on-voice but complete and correct.

Question 13

Does AI Drafting cost extra?

Accepted Answer

There's no free plan, but the 7-day free trial on Pro and Autopilot lets you try drafting and the inline tools. Full voice-matched drafting in the voice you set in your Personal Context, the context engine that grounds it, and the complete inline toolset come with Pro at $12.42/month. Autopilot at $20.75 adds bounded autonomous send of those drafts within your guardrails. On Autopilot and Team, BYOK lets you run the AI on your own provider key with no metered caps.

Question 14

Does it match different tones for different people?

Accepted Answer

Yes — voice-matching is per relationship, not one-size-fits-all. Set your default tone in your Personal Context, then use client context profiles and agent-behaviour rules to set standing tone and instructions per contact or domain — formal with a client, direct with a cofounder — without switching anything each time. So the same "Draft in my voice" action produces a polished, careful reply to your investor and a quick, casual one to your teammate — automatically.

Question 15

Can I edit or reject what the AI writes?

Accepted Answer

Always — you're never handed a take-it-or-leave-it draft. Every reply lands as editable text you can rewrite freely or reshape with one-click inline tools. Reject it entirely and write your own; the AI never sends on its own in Copilot, and in Autopilot a send-delay window lets you pull anything back before it leaves. Drafting is there to remove the blank page and the busywork, not to take the keyboard away from you.

Replies that sound like you wrote them.

Everything this does for you.

From inbox to done, step by step.

The payoff.

Where it earns its keep.

Good to know.

Related documentation.

Your voice, at the speed of one click.