Speak into the world · get an answer back

The AI that answers instantly in your ear.

  • Questions
  • Dictionary
  • Vicinity Translation
  • Two-way Language Translation
  • Translate with Someone Nearby
  • Job Help
  • Personal Assistant
  • Flirt
  • AI Meeting
  • Home Control
  • Custom Conversation
  • Souls

Reverbot listens with you and responds through your earbuds — questions, Dictionary, your job's documentation, real-time translation of the language being spoken around you, two-way conversations with someone who doesn’t share your language — or, if they have Reverbot too, a conversation where you each hear the other in your own earbuds — and what to say next when you go blank.

Coming soon on theApp Store Coming soon onGoogle Play

Get an email the day it's out →

iPhone & iPad first · Android to follow

Listening
How it works

Zero-friction, from pocket to answer

Reverbot is built for the moments your hands are busy and your ears are free — driving, working, walking, mid-conversation.

01

Tap once

Start a session and put your phone away. Reverbot keeps listening through your earbuds — even when the screen locks.

02

Just talk

Speak naturally, flat delivery and all. Reverbot detects real, answerable questions on its own and ignores small talk like “how are you”.

03

Hear the answer

Answers stream to your ear within a beat of you pausing. Concise by default, extended when you want more.

Powered by Jev

It knows when you’ve finished talking

Every voice assistant has to decide when you’ve stopped. Most just wait out a stretch of silence — too long and every answer drags, too short and it cuts in when you pause for breath. Reverbot uses Jev, TypeSafe’s split-second decision model, to read your words the moment you pause and judge whether you’re done — and to tell a soul when a little “mm-hm” fits.

Finished

No long wait

When Jev can tell you’ve finished, Reverbot answers at once at the default setting — or after exactly the wait you’ve set, if it’s longer.

Not yet

No cutting you off

Stop mid-sentence — “what’s the capital of…” — and Jev reads it as unfinished. Reverbot gives you a second more for the rest instead of answering half a question.

Ahead

The answer’s already coming

On Reverbot’s own AI, your question goes out at the first short pause, while Jev is still deciding, so the answer is already being written as Reverbot makes sure you’re done. Keep talking and it’s thrown away unheard.

Jev is an AI model made by TypeSafe AI, Inc. Reverbot is not affiliated with, sponsored or endorsed by TypeSafe. Jev sees only the text of what you say — never audio — and only after you agree; it goes to TypeSafe under TypeSafe’s own terms and privacy policy, Reverbot’s server doesn’t keep it, and it is never used on phone calls. The answer itself takes as long as your AI takes. Privacy Policy

Thirteen modes

One listener, thirteen superpowers

Pick an automatic mode for the moment. On the main screen they sit in three rows: Questions, Dictionary and AI Meeting; Translate Vicinity, Translate 2-way and Flirt; then Voice Changer, not a listening mode but a switch that changes how you sound whichever mode is running, and Custom Conversation — a conversation you write yourself, or a source that answers from your own documents — always there, with a button name of up to 15 characters if you want one. Turn on the wake-phrase modes in Settings → Wake Phrases, and they listen alongside whatever mode is running. One of the translation modes is for when the other person has Reverbot too, and lives behind its own button. And two switches ride along with the modes instead of replacing them: Answer questions too answers anybody's question out loud while a meeting, a translation or a coached conversation keeps running, and Translate too translates the room during an AI Meeting. And Souls — personalities you create, who remember you — live behind a button of their own and pick up whenever you call them by name.

Questions

automatic

Ask anything, any way. Reverbot spots genuine questions in natural speech — no question-y tone required — and answers only when there's a real answer.

“what's the capital of mongolia” · “who painted the mona lisa”
Read more →

Dictionary

automatic

Say a word, hear its top definition. Give it a sentence and you hear the meaning that fits — or ask how to spell it, where it comes from, or another word for it.

“ephemeral” · “what does ‘run’ mean in ‘I want to run to the store’”
Read more →

Vicinity Language Translation

automatic

Everyone around you arrives in your ear in your own language, while they're still talking. You never choose their language, or how many are in the room — it works that out from what it hears, every time, whatever they switch to.

“¿Dónde está la estación?” → you hear “where is the station?”
Read more →

Two-way Language Translation

automatic

Have an actual conversation. They speak and you hear it in your language; you answer, and the phone says it out loud in theirs — so you never have to pronounce a word. No beep to wait for, nothing to tap: it works out who spoke, and their language, on its own.

they say “¿Dónde está la estación?” → you hear it · you say “near the museum” → the phone says “cerca del museo” out loud
Read more →

Translate with Someone Nearby

both phones

They have Reverbot too? Then nobody holds up a phone. Your phones find each other over Wi-Fi and Bluetooth, and each of you just talks and hears the other in your own language, in your own earbuds — both directions, no turns. Works for a table of up to eight, each hearing everyone in their own language.

you say “where's a good place for dinner?” → they hear “¿dónde hay un buen sitio para cenar?” · they answer → you hear it in English
Read more →

Job

“hey job”

Upload your manuals, policies, walkthroughs — whole books if you want — and ask questions answered from your documents. Keep separate workspaces per job title, and switch which one is live.

“hey job … what's the torque spec for the head bolts”
Read more →

Personal Assistant

“hey assistant”

Tasks, reminders, calendar events, grocery lists, notes, projects — even drafting texts and emails. Every change confirms out loud, and a quick “undo” reverses it.

“hey assistant … remind me to call mom at 6” · “undo”
Read more →

Flirt

automatic

Never go blank mid-conversation. It suggests your next line — either the exact words to say or a nudge you put your own way — at whichever level of nerve you pick: Sincere, Charming or Daring.

“ask about the trip she mentioned — then tell her your worst travel story”
Read more →

Custom Conversation

you write it

Write the room yourself — Job Interview, Rent negotiation, Dinner with her parents. You set out who the other person is, who you are in that room, what a good outcome looks like, and which documents matter — a CV, the job ad, a page of notes. Or make it an Information Source that answers only from your own documents, if you choose. Its button is always on the main screen, next to Voice Changer, named Custom Conversation or anything you like up to 15 characters.

“they asked for a weakness — name a real one, then what you changed about it”
Read more →

Voice Changer

free effects

Sound like a robot, a giant, an alien or an old radio. More than twenty free effects run on your phone, with sliders for pitch, echo and more, plus a soundboard. Switch it on or change it any time, even mid-call: it changes your voice on Reverbot calls, video calls and the recordings you make with it — never what Reverbot hears. Natural AI voices for your recordings use your own ElevenLabs, Respeecher or Resemble account.

Robot · Giant · Alien · Radio · Retro game · or save your own
Read more →

AI Meeting

everyone agrees first

Put the phone on the table and let it do the writing. Reverbot records the room, transcribes it, and then writes the report — the decisions, who owns what, and what was actually agreed — in the length you choose. It is kept in a history you can go back to, so nobody has to be the one taking notes instead of talking. Questions asked across the table can be answered in your ear as it runs, and a room that speaks more than one language can be translated — on everyone's own phone, in their own language.

recording starts only once everyone in the room has agreed
Read more →

Home Control

coming soon

Say it instead of reaching for it. Lights, plugs, locks, fans, blinds and thermostats on your Google Home, spoken to the same earbuds you already have in — no app to open, no phone to pick up, no wake word shouted across the room.

“turn off the kitchen lights” · “set the thermostat to 20”
Read more →

Souls

“hey Sally, you there?”

Personalities you create and talk to — out loud or by text, one at a time or up to four on one call. Give them a whole life story and their own AI, voice and face; only one that lives in your phone can help on it. They remember you, and grow the more you talk.

“hey Max, you there? … turn the flashlight on”
Read more →
Reverbot Chat

A private messenger, sealed end to end

Message friends one to one or in groups, send photos, voice notes and files, call them by voice or video, meet on video and share your screen — all encrypted on your own phone with the Signal protocol, the design Signal and WhatsApp use. Not even Reverbot can read a message you send or hear a call you make.

Unrecoverable

Deleted is gone for good

Delete a message, or let it disappear, and it’s erased from your iPhone’s storage — not hidden. Every time, Chat marks the spot: “Permanently erased · Unrecoverable”.

Encrypted

Calls stay between you

Voice and video calls, meetings and shared screens are encrypted on the phones in the call. Reverbot can’t hear or see them, and nobody is recorded without agreeing first.

Sealed

Nothing readable on our servers

Our server only carries sealed envelopes it can’t open, and deletes each one the moment it’s delivered. The keys that open your messages never leave your phone.

Disappearing

Messages that vanish

From 30 seconds to 4 weeks, for everyone in the chat — deleted from every phone when the time is up. Delete for everyone works too.

Private

Nothing left lying around

Chats sit in protected storage on your phone, never in iCloud or computer backups. Notifications just say “New message”, and the app switcher shows a lock.

Verified

Know who’s on the other end

Safety numbers let you check nobody is in between — and if someone’s changes, nothing goes to them until you accept it.

How Chat protects you →

Vicinity Language Translation

A room full of languages.
One of them is yours.

Four people, four languages, four different things said — and each one arrives in your ear in yours as they say it. Nobody told Reverbot what to expect, and it doesn’t matter how many languages are in the room.

Four people nearby speaking different languages, each translated as they speak Four people on the left each say something different in Spanish, Japanese, German and Arabic. A person in the middle wearing an earbud receives a running feed of those remarks in English on the right, one arriving after another. ¿Dónde está la estación? 次は何時ですか Der Zug hat Verspätung من هنا مباشرة Where is the station? What time is the next? The train is delayed Straight on from here ¿Dónde está la estación?次は何時ですかDer Zug hat Verspätungمن هنا مباشرةWhere is the station?What time is the next?The train is delayedStraight on from here
Not one phrase shown four ways — four separate remarks, each translated as it is spoken. Nothing configured, no languages picked, no limit on how many are talking.

Questions

They ask.
You already know.

Leave the mic on Room and Reverbot hears the question put to you. The answer lands in your ear before the pause gets awkward — and what they hear is you, answering.

Someone asks you a question and Reverbot answers it in your ear Two people face each other. One asks a question out loud; the phone in the other person's hand picks it up in room mode, the answer arrives in their earbud, and they say it back out loud. Room mic “what’s ethanol boil at?” 78°C, into your ear “seventy-eight, near enough.” Room mic“what’s ethanol boil at?”78°C, into your ear“seventy-eight, near enough.”
You hear it first. They hear you say it. Room mode is what makes the difference — Solo would filter their voice out entirely.

Dictionary

Read on.
Never stop to look it up.

Say the word out loud and keep reading. The meaning arrives in your ear before you've lost the sentence.

A person reading in bed says a word aloud and hears its definition Someone lies in bed under a lamp reading a book. A word lifts off the page, and its definition comes back into their ear. “ephemeral” lasting a very short time “ephemeral”lasting a very short time
Say it once, out loud, without putting the book down. Reverbot doesn't need a question — a single word is enough.

Two-way Language Translation the phone speaks for you

You answer.
The phone says it for you.

They speak and you hear it in your language. You reply in yours, and the phone says it out loud in theirs — on speakerphone, so they hear it even with your earbuds in. Nothing for you to pronounce.

A two-way exchange in four steps, with the phone speaking your reply They speak; you hear it in your language; you reply in yours; the phone says your reply out loud in theirs. them you THEY SAY “¿dónde está la estación?” 1 YOU HEAR, IN YOURS “where is the station?” 2 YOU REPLY, IN YOURS “near the museum” 3 THE PHONE SAYS IT ALOUD “cerca del museo” 4 themyouTHEY SAY“¿dónde está la estación?”1YOU HEAR, IN YOURS“where is the station?”2YOU REPLY, IN YOURS“near the museum”3THE PHONE SAYS IT ALOUD“cerca del museo”4
Four steps, and you only do two of them: listen, and reply in your own language. The phone handles the rest — on speakerphone, so they hear it even with your earbuds in.

Two-way Language Coaching you say it yourself

It teaches you the line.
You deliver it.

Turn off “Phone speaks your reply” and it changes shape: the sentence arrives in your ear in their language, you say it out loud, and Reverbot listens until you’re close. Say “repeat” to hear it again.

A two-way exchange in five steps, where you say the line yourself They speak; you hear it in yours; you reply in yours; it tells you how to say that in theirs; you say it out loud. them you THEY SAY “¿dónde está la estación?” 1 YOU HEAR, IN YOURS “where is the station?” 2 YOU REPLY, IN YOURS “near the museum” 3 IT TELLS YOU HOW TO SAY IT “cerca del museo” 4 YOU SAY IT OUT LOUD “cerca del museo” 5 themyouTHEY SAY“¿dónde está la estación?”1YOU HEAR, IN YOURS“where is the station?”2YOU REPLY, IN YOURS“near the museum”3IT TELLS YOU HOW TO SAY IT“cerca del museo”4YOU SAY IT OUT LOUD“cerca del museo”5
The same exchange, except step 4 hands the line to you. Say “repeat” as often as you like — it costs nothing. If it still can’t make out your pronunciation, it moves on rather than leaving you stuck.

Flirt word for word

It hands you the line.
You just say it.

Listens along with you, hears when they’ve finished, and gives you the next thing to say — word for word, if that’s what you want.

Flirt in exact-words mode: it hands you the line and you say it They say something; Reverbot tells their voice from yours and hears them finish; the exact line arrives in your ear; you say it word for word. them you THEY SAY “just back from Lisbon” 1 IT HEARS THEM FINISH their voice, not yours 2 IT GIVES YOU THE LINE “so what took you there?” 3 YOU SAY EXACTLY THAT “so what took you there?” 4 themyouTHEY SAY“just back from Lisbon”1IT HEARS THEM FINISHtheir voice, not yours2IT GIVES YOU THE LINE“so what took you there?”3YOU SAY EXACTLY THAT“so what took you there?”4
It tells their voice from yours by how each one reaches the microphone, and takes the turn when they stop — nothing to touch. The suggestion arrives in your ear, not on the screen.

Flirt Coaching the move, in your own words

Or it tells you what to do.
The words stay yours.

Same loop, different help: instead of a script it gives you the move — and you say it however you'd actually say it.

Flirt in coaching mode: it tells you what to do and you phrase it They say something; Reverbot tells their voice from yours and hears them finish; it tells you what to do rather than what to say; you put it in your own words. them you THEY SAY “just back from Lisbon” 1 IT HEARS THEM FINISH their voice, not yours 2 IT TELLS YOU WHAT TO DO get her talking about it 3 YOU PUT IT YOUR WAY “so why Lisbon?” 4 themyouTHEY SAY“just back from Lisbon”1IT HEARS THEM FINISHtheir voice, not yours2IT TELLS YOU WHAT TO DOget her talking about it3YOU PUT IT YOUR WAY“so why Lisbon?”4
Pick this one if repeating a line verbatim would sound nothing like you.

Sincere

Say what you mean. Warm and honest, with no tactics behind it.

Charming default

Sincere, plus real conversational craft — callbacks, light teasing, and openings they can run with.

Daring

Teases and challenges rather than compliments. Creates tension on purpose. Higher risk, higher reward.

Both styles run at whichever of the three you pick.

Advanced skill, running live

Not lines. Reverbot applies real pickup craft on every turn — and knows when to stop.

  • Reads their interest
  • Calibrates escalation
  • Callbacks
  • Teasing as an equal
  • Knows when to exit

No negging, no pickup lines, no pressuring past a soft no.

Job Help

Hands in the engine.
Manual in your ear.

Upload the documentation once. After that it answers from your manuals, not from the internet's guess.

A worker asks a question and hears the answer from their own manuals Someone works on a machine with both hands. They ask a question aloud; it travels to a stack of manuals and the exact figure comes back into their ear. your documents “hey job — torque spec, head bolts?” 65 lb-ft, then 90° turn your documents“hey job — torque spec, head bolts?”65 lb-ft, then 90° turn
It answers from the manuals you gave it. Separate workspaces per job, so the electrician's book never answers the plumber's question.

AI Meeting

Walk in with nothing.
Leave with the write-up.

Three people, a conference room, and not one notepad between them. Everybody actually takes part — because nobody is stuck being the one who writes it all down. And the quiet half of a meeting — the aside, the question you don't want to derail the room with — happens in the app instead of across four different messaging services.

Three people meet with no notepads; the report is emailed to all of them afterwards Three people sit around a conference table with a phone in the middle. Nobody is writing. Their conversation flows into the phone, and afterwards an envelope goes out to three inboxes carrying the finished report. Decisions · owners · next steps Sent to all 3 attendees No one took a note Decisions · owners · next stepsSent to all 3 attendeesNo one took a note
Recording starts only once everyone in the room has agreed — that step cannot be skipped.

1 · Everyone agrees

One tap on AI MEETING, and the room says yes before anything is recorded. Add the people who should get the report — name and email — and they are remembered for next time.

2 · Just have the meeting

Phone on the table, hands free, nobody scribbling. It transcribes as you go, so the person who would have been writing is arguing their point instead.

3 · Talk without interrupting

Tap anyone in the room to message them, or pull a few people into their own thread. And if someone is not looking at their phone, nudge them — your name and photo land on their screen. All of it over the same direct link between the phones: no server, nothing to sign up for, gone when the meeting is.

4 · It lands in every inbox

Turn on automatic analysis and email, and when the meeting stops the AI writes the report — what was decided, who owns what, what is still open — then sends it to every attendee, from Reverbot's own server. No share sheet, no forwarding, no "I'll type this up tonight". Leave them off and it shows you the cost first, then writes it when you tap Analyse.

Brief or detailed, your choice — and you can have the full transcript attached underneath the summary. Everything is kept in a meeting history on your phone. While it records, Answer questions too answers what the room asks and Translate too translates what it says, and both land in the transcript. How AI Meeting works →

Personal Assistant

Think of it once.
It’s already written down.

The things you remember with your hands full — and forget by the time they're free.

A person cooking says a task aloud and it appears in their phone Someone stirs a pan with both hands. They say a task aloud; it travels to a phone on the counter and appears as a checked-off list item. “hey assistant — add task: buy coffee” “hey assistant —add task: buy coffee”
Said once, with your hands in a pan. It lands in your Personal workspace inside Reverbot — and in the iPhone's own Reminders too, if you turn on Reminders sync in Settings.

Reverbot-to-Reverbot Calls app to app — no numbers

Call anyone on Reverbot.
Anywhere. No phone number.

Everything above is built for the room you are standing in. Calling reaches the people who are not in it: anyone with Reverbot, found by their username, over the internet — no phone number, no carrier, nothing per minute. The call goes phone to phone, encrypted, and no friend request stands in the way.

A Reverbot-to-Reverbot call between two people, phone to phone Two people, one phone each, joined by a line. You search a username; their phone rings; they answer or decline; the call runs between the two phones, encrypted. them has Reverbot too you YOU SEARCH a username, not a number 1 THEIR PHONE RINGS anywhere in the world 2 THEY ANSWER or decline — they see who 3 PHONE TO PHONE encrypted, nothing per minute 4 themhas Reverbot tooyouYOU SEARCHa username, not a number1THEIR PHONE RINGSanywhere in the world2THEY ANSWERor decline — they see who3PHONE TO PHONEencrypted, nothing per minute4
Both of you have Reverbot, and that is the whole requirement: no numbers to swap, no friend request first — search a username and call. Audio travels directly when the networks allow, otherwise through a relay that cannot decrypt it.
Job knowledge

Your entire job, on recall

Reverbot turns whatever you feed it into voice-searchable help pages — and tells you plainly when a file couldn't be read, so nothing silently goes missing.

Upload anything

PDFs, photos of paper docs, plain text, RTF, HTML. Scanned pages are read with on-device OCR, page by page.

PDFImages + OCRText / MarkdownRTFHTML500-page books

Parsed & indexed on your phone

Live progress per file — “page 47 of 213” — with a clear status when something can't be extracted, and a one-tap retry.

Ask with your voice

“Hey job” routes your question to the active workspace's documents. Answers cite the doc they came from — and if the docs don't cover it, Reverbot says so instead of guessing.

Uploads — Mechanic workspace

Engine_Service_Manual.pdf
642 pages · OCR
Ready
torque-chart-photo.heic
1 page · OCR 97%
Ready
Company_Policy_2026.pdf
extracting · page 47 of 213
Parsing…
old-scan-1998.pdf
3 pages unreadable · rest extracted
Partial
Souls

Someone to talk to, who remembers you

A soul is a personality you create — a friend, a mentor, a character, the IT person who sorts out your phone. Call one by name whenever Reverbot is listening and just talk, out loud or by text. Every conversation is kept on your phone, and they grow from each one.

Make them

Start from scratch, pick a kind of person — each comes out as someone new, with a look and a voice drawn at random — describe one in your own words, or generate one with a whole life — where they’re from, what kind of person, or completely random. Then shape everything: personality dials, their story, moments from their life, documents about them.

Personality dialsLife storyBiographies & documentsGenerate a whole life

Call them by name

“Hey Sally, you there?” — they pick up and you talk, with little “mm-hm”s in your ear while you do. Say “bye” to hang up. Mid-call, “invite Sam” brings another soul in — up to four on one call, one voice at a time.

A face and a voice of their own

Each soul has its own AI (Grok is recommended), its own voice and speed — even a voice you record, with the speaker’s permission, on your own ElevenLabs account — and its own look while it talks: a free illustrated or 3D face, lips, an orb or a waveform. Only a soul that lives in your phone makes pictures and videos for you, and only once you allow it and ask — non-explicit only, on your own provider account, with the cost stated first and nothing made until you confirm. Or make your own in Settings → Media library.

Talking facesRecorded voicesPictures & videos

They grow

After each conversation a soul remembers what it learned about you and what happened between you, keeps a journal, and slowly becomes more itself — never further than you allow. How readily one warms to you is yours to set, from Very easy to Very challenging: closeness comes only with time and repeated conversation, and it can fall when things go wrong. Only a soul that lives in your phone can text, email, call or set reminders for you, or be your assistant. One with a life of its own — on a farm, in a fantasy realm — stays in its story, so it can’t.

Your souls

Max · IT
full access · your assistant
Assistant
Sam · Best friend
“hey Sam, you there?”
On
Morgan · Mentor
“hey Morgan, you there?”
On

Souls are AI, and Reverbot says so. For adults 18 and over; not therapy — if you’re in crisis, call or text 988 (US & Canada), or find a helpline in your country at findahelpline.com. How Souls works →

Life Journal

Your stories, in your own voice

A private journal of your life — memories you tell out loud, write down, or gather from the photos, videos and documents you pick. It stays off until you turn it on in Settings → Life Journal, and nothing in it leaves your phone unless you choose to send it — an interview, an analysis, a draft or a Journal Soul uses the AI selected for it, and a backup is encrypted before it goes.

Off by default

Nothing until you say so

Turning it on reads nothing and records nothing by itself. You sign in, so the journal belongs to your account, and you allow each source yourself, one at a time.

Record

Tell it out loud

Tap Record a memory, give it a title and talk, for up to 30 minutes. Play it back, and turn it into text on your phone whenever you want the words.

Collect

Write, or bring things in

Write a memory, or add the photos and videos you choose from your library and documents from Files — Google Drive or Dropbox too, once they are set up in Files. Only what you select comes in.

Listener

“Hey Life Journal Listener”

Turn on its wake phrase, then say it while Reverbot is listening to start a private interview: a simple guide, or a soul you choose, asks about your life one question at a time, and what you say goes to that listener’s AI.

Journal Soul

A soul made from your story

Build a Journal Soul from the entries you approve, with a voice you choose and, if you like, your recorded face. It is an AI portrayal, clearly labeled, and talking to it never changes what your journal says.

Encrypted

Locked on your phone

The journal is encrypted on your iPhone. Back it up to Reverbot only when you choose: it is encrypted before it leaves, and the key stays with you, in a recovery kit you keep.

How Life Journal works →

Voice displays

Choose what the voice looks like

The circle in the middle of the main screen shows the voice as it speaks. Keep the Reverbot blue, or pick from more than a dozen displays — one for the whole app, and a different one for any feature or soul.

Light, glass and retro

A glowing orb, matte or shiny glass and an ion sphere, in the color you pick — or a wave, a voice-memo scroll, a retro equalizer, an oscilloscope, or the light bars of an 80s talking car.

Talking faces

Free illustrated and 3D faces, or just the lips, moving with the words as they’re spoken. Style an illustrated one in the Character studio: skin tone, hair, eyes, glasses and a decorative frame.

A real face, with permission

Personal face and lips make the recorded face of someone who agreed to it talk, free, on your phone. For a realistic face, use your own Anam account for a live face, billed by the second, or your own Simli account for a clip of each sentence.

Per feature, per soul

Set the app’s display in Settings → Center button appearance, then give any feature, Custom Conversation scenario or soul one of its own.

Hands-free by design

Built for ears, not screens

Custom wake phrases

“Hey job” and “hey assistant” out of the box — or change them to anything you like. Turn them on in Settings → Wake Phrases, and they listen alongside whatever mode is running.

Earbud controls

Use next-track to skip an answer: press twice on AirPods Pro, 3 or 4. Play/pause stops a running session. Start listening again in the app.

Shake to skip

Phone in your pocket? A quick shake cuts off the current answer and goes back to listening.

Solo & Room modes

Solo isolates your voice; Room picks up everyone — for group settings or across-the-table questions.

Concise or Extended

Two answer styles: a tight sentence or two for speed, or a few more when you want the context.

Say “undo”

Every change the assistant makes opens a spoken undo window. Say “undo” and it's reversed — task, reminder, event, list item.

Audio cues

Short beeps mark listening, success, undo and cancel — you always know where you stand without looking. Switch them off in Settings with Audio cues.

Screen-lock friendly

Lock your phone and keep going. Sessions run through your earbuds with lock-screen controls on tap.

Privacy

Listening, without the creepy part

An always-listening assistant only works if you can trust it. Reverbot is built so the sensitive parts never leave your control.

On-device speech recognition

Your voice is transcribed by iOS's on-device engine whenever the device supports it — audio isn't shipped to a server just to be understood. Translation and live rehearsals are the exceptions: their voice models receive the audio. Full detail.

Recording only with consent

Meeting, rehearsal, soul and saved-conversation transcripts stay on your phone until you delete them. A room’s audio is saved only after everyone in it has been told and has agreed: AI Meeting can keep the meeting's recording, encrypted on your phone, and you can switch it off; and a video meeting's host can record it. Otherwise audio is kept only when you record or import it yourself — a Life Journal memory, a Voice Changer recording, recordings in a soul file you import, or a face and its spoken permission for a talking face. A voice sample for a soul goes to your own ElevenLabs account, and Reverbot deletes its copy.

Encrypted at rest

Uploaded job documents are stored with iOS file protection, and your AI keys live in the system Keychain — never in plain files.

You hold the keys

Add your API key and Reverbot asks that provider for the answer directly, under your account and your terms. Without one the question goes through ours on the way to Google, which forwards it and returns the answer without writing it down. Either way, the check for when you've finished speaking, translation and the default voice go through Reverbot's server, and calls are set up there. Full detail.

Bring your own AI

Your assistant, your model

Plug in the frontier model you already trust — paste a key once in Settings and you're done in about two minutes. Switch providers anytime. One of them costs nothing: Gemini is free every day, within the daily allowance Google sets on its free tier — their allowance to change.

Anthropic Claude
OpenAI
Google Gemini
xAI Grok
No special hardware

Reverbot works with your own headphones

There is nothing to buy. Reverbot speaks through whatever your phone already plays to — the AirPods in your pocket, the Bluetooth earbuds in your bag, an old wired pair, or the phone’s own speaker if you have nothing on you at all. Every mode on this page works completely, at full quality, through the headphones you already own.

AirPods

Any generation, Pro or not. Use next-track to skip: press twice on AirPods Pro, 3 or 4, or assign next-track to double-tap on older AirPods. Play/pause stops a running session.

Any Bluetooth pair

Sony, Bose, Beats, Samsung, the cheap set from the airport. Reverbot uses the standard Bluetooth audio profiles — if your phone can play music to it, Reverbot can talk through it.

Wired, too

USB‑C, Lightning, or a 3.5 mm jack on an adapter. Nothing to pair, nothing to charge, and no battery to run out halfway through a conversation.

Or nothing at all

Without headphones, answers and translations appear on screen. Turn on Speakerphone in Settings to hear them aloud. Two-way replies, souls and calls still speak.

Headphones keep spoken answers in your ears. The earpieces below change something different — whether anyone can see that you’re wearing anything at all. They’re third‑party hardware, they cost extra, and nothing in Reverbot needs them.

Wear it discreetly

Or wear one nobody can see

Visible earbuds change how a room reads you — and there are moments where you’d rather the help just not be part of the conversation. A live translation while someone is still speaking. A figure you need mid‑meeting. These are third‑party earpieces small enough to sit inside the ear canal.

360 ULTRA — single earpiece only

The 5 mm earpiece on its own — a spare, or a second one. No module, so it only works if you already have one. Up to 8 hours on an LR521/AG0 cell. Black. The retailer calls it their smallest non‑magnetic earpiece.

$69
View at 360invisible

360 NANO Spy Earpiece Set

The smallest of them at 2 mm, and the only one with no battery in your ear — sound reaches you by vibration instead. Up to 5 hours talk, 12+ standby on the module, which is worn on the neck with a loop antenna. Ships with eight earpieces, the module, a test earpiece and a USB cable.

$109
View at 360invisible

360 ULTRA Spy Earpiece Set

The standard set: the 5 mm earpiece plus a Bluetooth module worn on the neck with a loop antenna, which the retailer says keeps the signal stable however you move. Up to 8 hours on the earpiece, 5 on the module. Call button on the module.

$119
View at 360invisible

360 ULTRA Box Set

Same 5 mm earpiece, different module — the Box version has a built‑in antenna and isn’t worn around the neck, so nothing loops under your collar and it works with any clothing. Range about 30 cm from the earpiece, so keep it in a pocket.

$139$159
View at 360invisible

360 PREMIUM Box Set

Out of stock

Two earpiece sizes — PRO at 10 mm and ULTRA at 5 mm — with the no‑neck Box module and a detachable remote microphone with an answer button. Beige. Up to 8 hours on the earpiece, 5 on the module, 30 cm range.

$189$219
Check availability

360 PREMIUM Set

Out of stock

All three earpieces in one box — PRO 10 mm, ULTRA 5 mm, NANO 2 mm — plus a built‑in amplifier you switch on at the module, which is the point of this one: you can raise the volume for the smaller earpieces. Neck module, detachable remote mic with answer and signal buttons. Beige.

$199$229
Check availability

Sold by 360invisible, not by Reverbot — we don’t make, stock, or ship these, and we can’t support them. Prices and availability are what their site listed when this page was written and may have changed.

Get Reverbot

Put an answer in your ear instantly

Launching soon on iPhone and iPad. It will be free to download — bring your own AI key and start asking the air questions.

Coming soon on theApp Store Coming soon onGoogle Play

Get an email the day it's out →

iPhone & iPad first · Android to follow