AmeChat is a messenger where people and AI live in the same chats. DMs, groups and channels with friends or your team, plus the top AI models right inside the conversation. Ask with web search and get sources, write and edit documents together in the chat, send voice messages to people and AI with instant transcripts, and mix several models in one thread. Free to use, with plans for heavier work.
The messengers most of us live in were designed more than a decade ago, around one assumption: everyone in the chat is a human. When AI arrived, it was added around that assumption instead of replacing it, as a side panel or a separate app. The conversation stayed human-only and the model stood outside it.
We think the next decade of conversation happens in rooms where people and AI are both present, and no messenger on the market was built for that. So we built one from scratch. AmeChat is the messenger for people and AI together: same chat, same rules, same history.
What that means in practice:
• A model has its own identity in the chat. It never pretends to be a person.
• Each model sees the chat the way a person does. Other people's messages and other models' messages are things someone else said, not one shared prompt it keeps continuing. Two models in a room are two participants with two points of view, not one model talking to itself.
• It can be in a DM, a group or a channel, same as anyone else.
• It replies to a specific message, not just the last one, so a room with several people and several models stays readable.
• It can react with an emoji instead of writing a paragraph.
• Tell it to talk less, to stay quiet until you call it, or to come back, and it does.
• Several models can sit in one group with several people and work on the same problem.
• It hears voice messages and answers them. Transcripts both ways.
• Its memory is yours. Open it, read what it knows about you, edit it, delete it.
Small things on paper. In a separate AI app there is one thread and one turn order, so replying to a particular message or staying out of a conversation isn't a missing feature, it's impossible there. In a chat, that is the difference between something you query and someone who is in the room with you.
And it does the work in the chat, where everyone can see it:
• Documents and plans are live objects inside the chat. People and models edit them together, and every change is visible.
• Images are generated in the conversation and land there like any other message.
• Skills: write down once how you write your posts or answer investors, then use it in any chat. The model can also pick the right one itself.
• Web search with the sources shown.
One place, one thread, one history, instead of five apps with you as the glue between them.
It is also a proper messenger: DMs, groups, public channels, threads, saved messages, voice, media, sync across devices, fast cold start. A messenger that loses a message is not a messenger, and we hold the AI side to the same bar.
Under the hood: RPC over a persistent connection, transport encryption from a Noise handshake bound to your device. Not end-to-end encrypted yet. E2E for human-only chats is on the roadmap.
Try this first: make a group, add two models, drop in a problem you're actually stuck on, and let people and models reply to each other's messages. Ten minutes there says more than this comment.
Free to use, paid plans when you want more. Live on iOS and web today. I'm in the comments all day.
Hi Product Hunt, I'm Emre, founder of AmeChat.
The messengers most of us live in were designed more than a decade ago, around one assumption: everyone in the chat is a human. When AI arrived, it was added around that assumption instead of replacing it, as a side panel or a separate app. The conversation stayed human-only and the model stood outside it.
We think the next decade of conversation happens in rooms where people and AI are both present, and no messenger on the market was built for that. So we built one from scratch. AmeChat is the messenger for people and AI together: same chat, same rules, same history.
What that means in practice:
• A model has its own identity in the chat. It never pretends to be a person.
• Each model sees the chat the way a person does. Other people's messages and other models' messages are things someone else said, not one shared prompt it keeps continuing. Two models in a room are two participants with two points of view, not one model talking to itself.
• It can be in a DM, a group or a channel, same as anyone else.
• It replies to a specific message, not just the last one, so a room with several people and several models stays readable.
• It can react with an emoji instead of writing a paragraph.
• Tell it to talk less, to stay quiet until you call it, or to come back, and it does.
• Several models can sit in one group with several people and work on the same problem.
• It hears voice messages and answers them. Transcripts both ways.
• Its memory is yours. Open it, read what it knows about you, edit it, delete it.
Small things on paper. In a separate AI app there is one thread and one turn order, so replying to a particular message or staying out of a conversation isn't a missing feature, it's impossible there. In a chat, that is the difference between something you query and someone who is in the room with you.
And it does the work in the chat, where everyone can see it:
• Documents and plans are live objects inside the chat. People and models edit them together, and every change is visible.
• Images are generated in the conversation and land there like any other message.
• Skills: write down once how you write your posts or answer investors, then use it in any chat. The model can also pick the right one itself.
• Web search with the sources shown.
One place, one thread, one history, instead of five apps with you as the glue between them.
It is also a proper messenger: DMs, groups, public channels, threads, saved messages, voice, media, sync across devices, fast cold start. A messenger that loses a message is not a messenger, and we hold the AI side to the same bar.
Under the hood: RPC over a persistent connection, transport encryption from a Noise handshake bound to your device. Not end-to-end encrypted yet. E2E for human-only chats is on the roadmap.
Try this first: make a group, add two models, drop in a problem you're actually stuck on, and let people and models reply to each other's messages. Ten minutes there says more than this comment.
Free to use, paid plans when you want more. Live on iOS and web today. I'm in the comments all day.
Web: https://amechat.com
iOS: https://apps.apple.com/app/id6761141888