AI Group Call 让用户输入目标后，立即加入与六位 AI 专家的实时语音讨论，自动记录并总结要点，适合需要集体智慧的用户。
热门评论
PH 用户
Hi Product Hunt, I am Tash, the maker.
AI Group Call started with a small frustration. I could ask one chatbot for advice, but the thing I actually wanted was the room: the meeting where a skeptic pokes a hole in your pitch, a strategist reframes it, and someone plays the customer who has to sign off.
So I built the room. You type a goal, and a few seconds later you are in a live voice call with six AI participants cast for that specific goal, each with a name, a role, and a personality. They speak one at a time, build on and disagree with each other, and go quiet the moment you start talking.
That last part was by far the hardest bit. Real barge-in over a phone speaker means the mic also hears the agents, so there is a lot of echo handling and raw audio voice detection under the hood to make interrupting feel like a real call instead of a walkie talkie. Getting that right is most of why this took as long as it did.
A few things people usually ask:
- Every participant is AI. There are no humans on your calls. - Every call is transcribed, can be summarised into key points and action items, and can be rejoined later with the same cast. - You can tap any agent mid call and rewrite their name, role, or personality. - Android is live today. iOS is built and waiting on submission.
Every new account gets a free minute, no card. After that it is monthly minute bundles starting at $4.99.
What I would genuinely like from today: throw a goal at it that I would never have thought of, and tell me where the cast falls flat. I am reading every comment.
PH 用户
Loved the idea Tash! need to try it out. Anyway wish you all the best here!
PH 用户
Thats cool, i have a cross functional AI team that im thinking of moving from text to audio. Did u use a framework for orchestration of the agents like crewai or langgraph? have a technical post on it? would love to see how u did it
PH 用户
When's the IOS app out please?
PH 用户
the barge-in echo problem is the real boss fight in voice, way harder than people expect until they hit it. are you doing echo cancellation on-device before it hits the model, or is the interrupt detection happening server side on the mixed stream
AI Group Call started with a small frustration. I could ask one chatbot for advice, but the thing I actually wanted was the room: the meeting where a skeptic pokes a hole in your pitch, a strategist reframes it, and someone plays the customer who has to sign off.
So I built the room. You type a goal, and a few seconds later you are in a live voice call with six AI participants cast for that specific goal, each with a name, a role, and a personality. They speak one at a time, build on and disagree with each other, and go quiet the moment you start talking.
That last part was by far the hardest bit. Real barge-in over a phone speaker means the mic also hears the agents, so there is a lot of echo handling and raw audio voice detection under the hood to make interrupting feel like a real call instead of a walkie talkie. Getting that right is most of why this took as long as it did.
A few things people usually ask:
- Every participant is AI. There are no humans on your calls.
- Every call is transcribed, can be summarised into key points and action items, and can be rejoined later with the same cast.
- You can tap any agent mid call and rewrite their name, role, or personality.
- Android is live today. iOS is built and waiting on submission.
Every new account gets a free minute, no card. After that it is monthly minute bundles starting at $4.99.
What I would genuinely like from today: throw a goal at it that I would never have thought of, and tell me where the cast falls flat. I am reading every comment.