I’d like to share an idea (and I hope I’m not just reinventing the wheel) that I believe could significantly enhance the daily experience of smartphone users. Specifically, I’m envisioning an intelligent voicemail system powered by a GPT model, capable of answering and screening incoming calls, analyzing requests, and—when appropriate—politely, humorously, or more assertively declining them.
In addition, this system could feature a fully customizable virtual assistant, configured through a dedicated prompt. This would let users define the exact interaction style and call-handling rules. While solutions like Google Duplex already tackle some aspects of phone interactions, I haven’t come across a dedicated application focused solely on advanced call filtering.
I’d love to hear the community’s thoughts on whether this could be a valuable addition to our communication toolkit. Have you seen anything similar, or do you have ideas on how to bring this to life?
“Hi, this is virtual Bob. Real Bob isn’t taking calls right now, or flat-out didn’t pick up on you. You’d better make a useful message for Bob for me to pass along in our chat, or I’m dumping it in the garbage.”
Sounds like a good way to spend $2 of voice AI per caller if they are polite enough to not chat about the weather and the AI is smart enough to call a hang up function after a voice trigger since it only acts on voice detection.
You would also have to be “on then network”, able to take voicemail redirections from carriers to alternate numbers you provide to really integrate. See Google Voice where your Google voice phone number can also handle voice mail for your cel phone as such an integration, forwarding after ring count.
Thx for bringing up the $2/call issue. Let me clarify my POV (disclaimer: unless, as a noob, I’m talking nonsense here—totally possible, lol):
No external APIs: Not suggesting 3rd-party svc, but a simple upgrade to OpenAI’s app. It could use the same voice features already in the Plus/Pro plan.
Admin perms: Once perms for contacts/calls are granted, the app could handle calls directly on the phone, routing audio like it does now for ChatGPT’s voice—but w/o any external svc.
Same voice interaction: The call would work like a regular GPT convo: caller talks, app processes via GPT, and replies. To avoid TL;DR convos, a max duration could be set in the prompt.
No extra $: No need for $2/call. Processing would rely on the same infra already included in Plus/Pro, so no change to monthly costs.
Hope this makes sense and shows why I think no extra charge is needed. If OpenAI decides to roll this out, the tech/cost principles should stay within existing sub limits.
Thx again for the great topic! (And if my take is BS, feel free to call me out—I’d probs deserve it!)