I’d love to continue using VoIP.ms, but if the platform doesn’t integrate with AI or support WebSocket (the popular technology stack for voice AI), the road ahead for voip.ms will be challenging.
VoIP.ms sets itself apart in a crowded market with robust PBX-like features (even if it doesn’t position itself as a full PBX, just a SIP trunk). However, as voice AI tools advance, a platform lacking native WebSocket support will face significant hurdles.
Ultimately, with AI, many traditional calling features become unnecessary. For example, I developed a voice AI application whose selling point is “No more voicemail, no more call directories” — two major PBX features that AI can render obsolete.
Are you also including the ability to create voice messages and IVR using only synthetic voices based on a text instead of always having to provide a fully recorded audio file?
I have made a Twilio and ElevenLabs voice ai application. I made it modular and vendor neutral (as much as possible). It’s really powerful. This is the future. ElevenLabs can now import none Twilio number. But for VoIP.ms, if at the end, it becomes really just another sip trunk, I really don’t think that’s a bright future.
Nice thread and good idea to have that integrated, somehow.
I would even want that at home when people call too late in the day. AI could answer, take the message, or decide it is worth waking me up and ring the phones.
this is a service i am developing now. one major challenge is for regular cell phone, at least on ios, there is no conditional forward. i am developing needed features first anyway.
Entirely conversational AI. Who still needs VM or IVR when calls can be answered 24/7.
I consider your proposal to be an IVR … an advanced IVR that accepts voice prompts instead of “press one for sales, press two for support”. An advanced AI would answer the phone with an open ended question. “Thank you for calling XYZ, how can I help you today?” and listen for keyword that the system has learned either through scripting or machine learning.
The problem with AI is the compute overhead needed to make it work. Voice recognition and text to speech would need to be excellent (not just good) to make people comfortable using the service. Speaking to the caller in a robotic voice and/or not understanding the caller’s response could make an AI-IVR more annoying to callers than a normal IVR.
I would expect the work to be done by a cloud server or a system on site at the business using the service, voip.ms would only be for “dial tone”. It would connect to the service just like a PBX and the AI endpoint should be able to answer, transfer and manage the call using standard SIP protocols.
I hope you find success with your work. Perhaps some day I will talk to your AI and not know it is AI.
Considering an IVR is an interactive voice response … even if people moved to your AI they would be using a form of interactive voice response. (The interaction being all voice and no number pad with potentially personalized voice responses)
I have not met a voice AI that I didn’t know was an AI within a few responses.
What call control are you looking for that cannot be done via a SIP interface?
Agreed, but we still don’t have the ability to transfer calls offsite or within Voip.ms without having to beg them to enable the feature that is hidden on the back end. (same with BLF) these are commonly used PBX features that have been around for 20+ years. I absolutely agree with you for the use of AI.. but I can only hope that they read this and actually do some of the tech/feature upgrades that we’ve been asking for, literally for years now.
