Atona/How it works/Email
Email

Putting your inbox on autopilot.

Atona reads, replies, and triages your inbox — auto-sending when confident, holding drafts when not, and keeping anything sensitive away from AI entirely.

In this guide: how Atona decides what to send, how to teach your voice, what counts as sensitive, and how to dial confidence up or down.

What it does

Email is where most of your day disappears. Atona handles the routine: confirming meetings, answering simple questions, sending the "got it" reply, archiving the noise — and asks you about anything it isn't sure of.

When you want a clean slate, tell Atona to sweep your inbox. It reads every unread email, classifies what needs a reply, drafts the ones it can handle, and queues the rest for archive or delete. You review and approve — inbox zero without the Sunday-night dread.

You choose how replies go out. Atona can draft in your voice and send from your personal Gmail once you approve — recipients never know an AI was involved. Or Atona can respond autonomously from its own dedicated address, handling the conversation end to end.

Your assistant's own address

Every user gets a dedicated email — something like [email protected]. It's not just for outbound. Anyone can write to it directly.

Contacts, vendors, recruiters — they email your assistant, and Atona handles the conversation end to end. It checks your calendar before proposing times, pulls context from past conversations, follows your standing instructions. You're CC'd on everything, so you see what went out without having to do anything.

You can also route outbound through it. Tell Atona "send that from your address, not mine" and your personal inbox stays clean — no back-and-forth threads cluttering your Gmail.

Try saying

From any channel — chat, text, voice — these all just work.

"

Reply to Sarah's email — say yes to Tuesday.

"

Draft a polite no to that recruiter.

"

Archive anything from newsletters I haven't opened in a month.

"

What's waiting on me right now?

"

Sweep my inbox — get me to zero.

"

Follow up with James if he doesn't reply by Thursday.

"

Send that from your address, not mine.

How it works

1

Atona reads new mail

Every minute, in your name, with full context — your calendar, your contacts, the conversation history. Sensitive messages are recognized and held back before any AI sees them.

2

It decides what to do

Auto-reply, draft for review, label, or escalate. The decision depends on the confidence threshold you set and the rules you've taught it.

3

It writes the reply in your voice

Tone matched to the relationship, context pulled from your memory, calendar checked for any time it proposes. Depending on your settings, the reply goes from your Gmail (with your approval) or autonomously from your assistant's address.

4

It follows through

Deadlines mentioned in emails become reminders. Promises Atona made — "I'll send that over" — become tracked tasks. If someone doesn't reply, Atona nudges at the right time.

5

You see what happened

Every action shows up in your morning brief. Anything that needs your call appears at the top. Review what Atona handled, approve or reject drafts, batch-manage the rest.

What you control

Send mode

Start with drafts — review everything before it goes out. Once you trust what you're seeing, flip to autopilot and let Atona send on its own.

Per-contact rules

"Always auto-reply to Sarah." "Always draft to my CEO." "Never touch my therapist."

Sensitive list

Add keywords or domains Atona must never reply to or summarize.

Quiet hours

Hold all auto-replies overnight. Atona sends them in the morning instead.

Inbox review

See every decision Atona made — what it replied to, what it flagged, what it wants to clean up. Approve or reject drafts. Archive or delete low-value email in bulk.

What Atona will not do

Common questions

What counts as "sensitive"?

Anything from a bank, medical provider, legal counsel, or government agency. Anything containing personal identifiers like account numbers, social security numbers, or health information. These are recognized and held back before any AI sees them.

Can people email Atona directly?

Yes. Your assistant has its own address. Anyone — contacts, vendors, recruiters — can write to it, and Atona responds with full context. You're CC'd on everything. You can point people to it yourself: "Email my assistant and she'll find a time."

Will it reply from my Gmail?

It can. Atona drafts in your voice and holds it for your approval — once you approve, it sends from your Gmail as you. Recipients see your name, your address, your tone. Alternatively, Atona can handle conversations autonomously from its own assistant address.

How does it learn my voice?

From how you've written in the past, plus standing preferences you've taught it. You can also override the tone for any specific reply.

Can I undo a reply?

Yes. For any reply Atona sent, you can recall, revise, or have it apologize on your behalf.

What about spam?

Spam never reaches Atona. It stays in your spam folder where it belongs.

Related guides

For multi-step sequences triggered by email — onboarding, follow-up campaigns, deal tracking — see Workflows.

Ready to try Atona?

Free during private beta.

Get early access