At Rime, we're passionate about crafting AI voices that don't just sound human, they feel human. And we deeply believe that improving voice AI authenticity leads to better business outcomes from increased conversion and retention to higher sales and beyond.
Our collaboration with Phonely, a trailblazer in AI-powered phone support, exemplifies how two companies can unite their strengths to redefine what's possible in voice technology.
🚀 A Meeting of Minds
Recently, our CEO, Lily Clifford, sat down with Phonely's CEO, Will Bodewes, for an in-depth conversation about the evolving landscape of voice AI. The dialogue was rich with insights, humor, and a shared vision for the future.
Rather than a rigid Q&A or polished soundbites, this was a candid, behind-the-scenes conversation between two people who have spent the last few years deep in the weeds of building real voice products. What emerged was an honest exchange about customer pain points, technical challenges, and the surprising realities of getting AI systems to actually do what businesses need.
Lily Clifford:
"So what are you seeing from customers as far as boots on the ground? What do they really care about?"
Will Bodewes:
"One area we're seeing these voice agents successfully deployed at scale is in lead qualification, like for example Medicare. You know, any place where you dial up someone on the phone, do a little qualification, and then pass them over to a sales rep or book an appointment, we're seeing a lot of big wins on that."
Lily Clifford:
"We see that too. The problem for outbound lead qualification is a very constrained one. And those constrained problems are so imminently automatable, therefore really valuable."
🧠 Tackling the Nuances of Voice AI
One of the challenges in voice AI is ensuring accurate pronunciation, especially for unique names and places.
Will Bodewes:
"One of our big customers in Australia runs into pronunciation issues constantly. Town names like Walbundrie or Manangatang or Moombooldool just get butchered. That erodes trust."
Lily Clifford:
"Exactly. If your model can’t pronounce ‘chalupa,’ you’re not building phone ordering for Taco Bell. These aren’t just edge cases. They’re real blockers."
Will Bodewes:
"What’s great is that you let us input custom pronunciations, record them, and guide the model output. That kind of control is huge for us and our customers."
Lily Clifford:
"We even have a free coverage check API. It flags any word we had to guess on, so your team can fix it before a customer hears it mispronounced."
To address this, Rime offers tools that allow voice AI developers to identify and correct pronunciation issues proactively.
🛠️ From Demos to Deployment
While creating a demo is straightforward, deploying a reliable voice AI system in production is a different ballgame.
Lily Clifford:
"Sometimes people have this idea that voice AI is good for demos but not good for production, but that’s just simply not true."
Will Bodewes:
"Yeah, it’s really easy to stand up some little demo that just strings together some APIs and chats back and forth and be like, oh, I could do everything, right?"
Lily Clifford:
"Exactly. Talking with people who have been building really impactful voice applications for a long time, they know how hard it is. LLMs solve some problems, but they create other problems too. I've never been one to be pessimistic about the wave of end-to-end technologies, but the people who expect that in five years you're going to be able to have a single API to make every kind of business call that has ever and will ever be made are in for a rude awakening."
Will Bodewes:
"Yeah, I totally agree. What we tell our customers is you’re the last mile for brand and content, and we’re the last mile for getting the voice application to work in production. Most teams don’t want to build the whole thing, they just want an agent that works. They want it to make calls, send emails, texts, integrate with the CRM, run A/B tests, etc.. But making all of that seamless takes way more work than you’d think. When I started the company two years ago, I thought I could just connect a voice to an LLM and a transcription service and be done. But we’re still solving all the little problems that make it actually sound human. Customers really care about this."
🎯 The Power of Collaboration
Our partnership with Phonely is more than just a technical integration; it's a fusion of shared values and complementary expertise. Together, we're committed to pushing the boundaries of what's possible in voice AI.
Lily Clifford:
"This is a hard problem, and at the same time, it’s solvable. Maybe not in principle, but example by example. That’s what it takes. Hard work and attention to detail."
Will Bodewes:
"And you’ve already done that backend work, which is awesome. Now we’re looking at how to bring that to the front, so our customers can actually use it, fix things fast, and make the experience feel real."
Lily Clifford:
"Right. We just want to make it easier for people to build really impactful voice applications by focusing on the modeling layer."
🌟 Looking Ahead
The future of voice AI is bright, and collaborations like ours are lighting the way. By combining Rime's advanced speech synthesis capabilities with Phonely's robust AI phone systems, we're setting new standards for customer interactions.
Stay tuned for more updates as we continue to innovate and redefine the voice AI landscape together!
