Automate the busywork
of being you.
ALTRmode learns how you write to each person and drafts the reply you'd send, in email, Slack, Teams and the tools you work in. When someone asks for something, AI agents called GENTs do the work, and it comes back as your reply.
Got an invite? Create your account · Already in? Sign in
Stage: contract sent · $48k · close date Oct 15
Found it in the session middleware. Fix pushed to claude/login-500. Open a pull request into main?
set per thread
Everything is off until you switch a thread on: a person, a group, a channel, an email thread or a project. Off means nothing is read, stored or learned. GENTs on Ask to change look things up on their own and ask before changing anything.
what it takes off your plate
Replies in your voice, and the work behind them.
Replies that sound like you
- Learns each relationship from its history: tone, length, language, punctuation, emoji, how fast you answer.
- Writes the reply you'd write, or none when you wouldn't answer. A reaction when that's what you'd do.
- Doesn't know something only you know? It asks you instead of guessing, and you can answer in the app, by email or from the notification.
- Review, or Automatic in the threads and hours you choose. Sensitive topics never go out on their own.
GENTs that do the asks
- Spots when a message asks for something: a ticket's status, a deal update, a meeting, a refund, a bug fix. Screenshots and links included.
- The GENT Lead works out what's really wanted and hands each part to the GENTs that fit, in your own tools, other AI agents or MCP servers.
- Looks things up on its own; changes (a refund, a record, a pull request) wait for your OK unless you allow them.
- Watch it work and steer it as it goes. The result comes back as your reply, in that thread's voice.
how it works
Three steps, at your own pace.
- 01
Connect where you talk
Email, Slack, Teams, Google Chat, help desks, project tools and more, each with its own sign-in. Every thread starts off.
- 02
Switch a thread on
A person, a group, a channel, an email thread or a project. Only ones you've been writing in lately can be switched on, so strangers never get a reply. It learns how you are with that person from the history. Connect your tools so GENTs can do the work too.
- 03
Review, then trust
Drafts wait for your tap, and your edits teach it. Automatic unlocks once you've sent a few messages in that thread, and you choose where to use it, for replies, for work, or both.
works where you already are
Email, chat, support, reviews and projects.
Gmail, Outlook and any other mailbox, shared addresses like support@ included. Replies go in the same thread, with your signature.
Chat apps
Slack, Microsoft Teams, Google Chat, Discord, Facebook Messenger and Instagram: direct messages, groups and channels, in the tone you use with each.
Help desks and reviews
Zendesk, Freshdesk, Intercom, Help Scout and Front tickets, plus Google, Trustpilot, App Store and Google Play reviews.
Projects and issues
Comments in Asana, Trello, monday.com, GitHub, Jira and Linear, then the task marked done, handed back or assigned, as you choose.
one voice, many registers
You don't write to your CTO like you write to your mom. Neither does it.
Style is learned per relationship, not globally. Same question, two people, two replies that each sound like you. Want something done one way everywhere? Add a rule in Voice (“sign off with just my first name”) and it wins.
ai agents for every ask
"Can you check…" becomes the work, then the reply.
When a message asks for something, the GENT Lead hands each part to the GENTs that fit, in your own tools. Changes wait for your OK, and the result comes back as your reply.
what that looks like
Every tool GENTs work inHubSpot
Looks up the Globex deal and logs a note on it.
Shopify
Finds order #1042 and its tracking, and refunds it once you OK it.
Claude Code
Finds what's behind the login errors and pushes a fix for your review.
Jira
Says where the ticket stands, or opens one for the bug.
Calendly
Sends a one-time booking link for the right kind of meeting.
Stripe
Finds the double charge and refunds it after your OK.
AWS
Answers “is it down?” from your logs, metrics and alarms.
Notion
Finds the answer in your pages, or writes the notes up as a new one.
beyond replies
Ask it, teach it, and let it keep you on top of things.
Answering is where it starts. The rest of the app helps you run your day around your messages.
CTRL
Ask in plain words: “what's waiting for me?”, “set Dan to Review”, “refund order #1042”. It hands work to your GENTs and runs routines on a schedule.
TUNE
Teach it once. Your Playbook says how you handle situations, Voice how you write, and Life what it should remember about you.
Calendar and to-dos
All your calendars in one list, changed by telling it (“dentist Thursday at 4”). What people ask you to do lands on your to-dos, with reminders.
Spaces and Focus
Sort threads into Work and Personal, then Focus on one for a few hours while the other waits quietly. Quiet threads rest and wake by themselves.
Act from the notification
Send a draft, answer its question, or allow a GENT's step right from the notification, without opening the app.
In step with your apps
Read here, read there. An email you're done with is archived in Gmail or Outlook, and typos you make are fixed before your message goes.
security
Strangers get nothing. Scams get a warning, not a reply.
An assistant that answers for you is only safe if it can't be talked into things. So it only works where you really talk, checks every message first, never spends your money, and says when a reply was automatic.
Only people you really talk to
A thread can only be switched on once you've been writing in it lately. A cold email, a new number or a spammer has no history with you, so nothing they send is read, stored or answered.
Every message checked first
Phishing, scams, dangerous files and spam are caught before anything is written, even from a contact whose account was hacked. You get a warning and a one-tap Report, never a reply.
It never buys anything
An offer to buy, order or subscribe is never accepted, never starts a GENT and never gets a reply sent on its own. Spending money is always your call.
Honest about automatic replies
Replies it sends on its own end with a short line saying they were automatic, linking to a page that explains it. Always on in the EU and EEA; elsewhere it's your choice.
built to be trusted with your accounts
Your accounts, on your terms.
Off by default
Per thread, and only ones with recent history between you can be switched on. Off means no import, no storage, no model calls.
Review, always at the start
Every reply waits for you until you decide otherwise, thread by thread.
Human timing
Automatic replies wait until they've stopped typing and go out at a natural pace, only in the hours you allow.
It asks when it should
Anything it can't know, or that needs you, comes to you by push or email, and you can answer right there.
Encrypted, with two-step sign-in
Your integrations and what it learns are encrypted. Every account signs in with an authenticator app.
No training on your data
Your messages improve your own replies only. Never sold, never used for ads or to train AI models.
Your tools, your say
GENTs act with your own sign-ins, only where you let them work (on GitHub, only the repositories you tick). Changes wait for your OK unless you allow them.
Delete means delete
A thread switched off for 72 hours has its messages and what it learned deleted. Delete your account and your data goes at once.
Your voice stays yours
What it learns from your account stays in your account. Nobody else sees your messages or your voice.
access
A private beta, by application.
We're letting people in a few at a time so every tester gets our attention. There's nothing to buy during the beta.
What testers get
free while in beta- Replies in your voice, Review or Automatic
- Work handed to GENTs in your tools
- A 20-minute setup call with us, if you'd like one
- A direct line to the team building it
What we ask
- Use it on real threads, every day
- Tell us what works and what doesn't, in the testers' group
- Keep what you see to yourself while it's in beta
faq
Questions people ask before switching a thread on
+What is ALTRmode?
An AI that answers your messages for you, in your own voice, wherever you talk: email, Slack, Teams and other chat apps, help desks, review sites and project tools. It learns how you write to each person, drafts the reply you'd send (or sends it on its own where you allow it), asks you when it doesn't know something, and hands real requests to AI agents, called GENTs, that do the work in your own tools.
+How is this different from an auto-reply?
An auto-reply sends the same message to everyone. ALTRmode writes a real answer to each message, in the tone you use with that person, or no answer when you wouldn't reply. Every thread starts on Review, so nothing goes out until you tap Send.
+Which apps does it work with?
Email (Gmail, Outlook and any other mailbox), Slack, Microsoft Teams, Google Chat, Discord, Facebook Messenger and Instagram, project tools (Asana, Trello, monday.com, GitHub, Jira, Linear), help desks (Zendesk, Freshdesk, Intercom, Help Scout, Front) and review sites (Google, Trustpilot, App Store, Google Play). Each one connects with its own sign-in. The full list is at altrmode.com/integrations.
+How do I get in?
Apply for the beta at altrmode.com/beta. We pick a small group for each round and email you when your spot is ready.
+What does it cost?
Nothing during the private beta.
+Will people know it's AI?
Replies it sends on its own end with a short line saying they were automatic, with a link to a page that explains it. Replies you approve or write yourself never carry it. You can turn the line off, except in the EU and EEA, where the law asks for it; with it off, telling people where the law requires it is up to you.
+What does it read?
Only threads you switch on (a person, a group, a channel, an email thread or a project), from the moment you switch them on, plus that thread's recent history to learn how you write there. Everything else is invisible to it. You can only switch on threads you've been writing in lately, so new senders never reach it.
+What about spam, phishing and scams?
Two layers. First, strangers never reach it: a thread can only be switched on once you've been writing in it, so cold emails, new numbers and spammers are never read or answered. Second, in the threads you did switch on, every message is checked before anything is written (a friend's account can be hacked too): phishing links and look-alike addresses, requests for passwords or login codes, gift-card and crypto scams, dangerous attachments and bulk spam. A flagged message gets no reply and starts no work. You see a warning with its links switched off, a one-tap Report that sends email to your spam folder, and Not spam if it got it wrong.
+Could it buy something, or say yes to an offer?
No. When a message offers something to buy, order or subscribe to, it never agrees, never starts a GENT for it, and any reply waits for you, even on Automatic. Spending money always needs your OK.
+Can it send something wrong?
Yes. It's AI. That's why every thread starts on Review, why Automatic only unlocks after you've sent a few messages there yourself, why sensitive topics and anything it isn't sure about never go out on their own, and why the Terms make you responsible for what's sent. Use Automatic only where a mistake is harmless.
+Which tools can GENTs work in?
Dozens, each with your own sign-in: Notion, Jira, HubSpot, Salesforce, Shopify, Stripe, Google and Outlook Calendar, Google Drive, OneDrive, SharePoint, Box, Zoom, QuickBooks, Xero, DocuSign, AWS, GitHub and more, plus MCP servers and other AI agents you connect, and a built-in research GENT that searches the web. For code, GitHub Copilot's coding agent runs on your own Copilot plan, and Claude Code only pushes to its own claude/… branches. Changes like opening a pull request wait for your OK unless you've set that thread to Automatic. The full list is at altrmode.com/integrations.
+Can I ask it to do things myself?
Yes, in CTRL. Say what you want in plain words: "what's waiting for me?", "switch on my chat with Mom", "open a Jira ticket for the login bug". It hands the work to your GENTs, asks before changing anything, and can run routines on a schedule ("every weekday at 8, tell me what's waiting").
+How do I stop it?
Switch a thread off (instant), switch ALTRmode off for a while in Settings (a holiday, a day off), disconnect the channel, or delete your account. A thread that stays off for 72 hours has its messages and what it learned there deleted; deleting your account deletes your data at once.
+Where is my data?
Encrypted in our US-hosted database, and messages are deleted 90 days after they were sent. To write and check replies, the messages needed go to our AI model providers (Anthropic, and OpenAI for some steps) under terms that forbid training on them. The Privacy Policy lists every sub-processor.
Switch on ALTRmode. Switch off the busywork.
Pick one thread and see if it sounds like you. Private beta, free while it lasts. Every thread starts on Review, and you can switch it off any time.