[0:05] **Me:** Good afternoon. It's such a pleasure to meet you all today. You have such

[0:06] **Me:** a beautiful office. Have you been here long?

[0:11] **Voice 2:** Thanks for coming in. I'm Monique. I run operations here.

[0:14] **Voice 2:** Super busy day, so I may duck in and out.

[0:18] **Voice 2:** Raj here owns the contact center tech stack.

[0:23] **Voice 2:** I, yeah, I care about uptime,

[0:23] **Voice 2:** headaches, and not blowing up our road map.

[0:28] **Voice 2:** And I'm Elena, CX and Insights. I'm listening for quality, customer

[0:31] **Voice 2:** impact and what this does to our metrics.

[0:37] **Me:** Great. And just to introduce myself, please,

[0:44] **Me:** for those of you I haven't already met, my name is Colin. I'm the newest account executive here at Rhyme. I'm someone who did the the the company, but I've actually been in a speech AI, gosh, almost

[0:46] **Me:** four years now starting with the company name

[0:51] **Me:** Deepgram. I was employee number one there. And, subsequently, at

[0:53] **Me:** Intelligent Voice and, and Gradium.

[0:56] **Me:** So very passionate about the space.

[0:59] **Me:** Jeez.

[1:04] **Me:** I guess we've done introductions from your side. So in terms of an agenda, you know, what I wanted

[1:07] **Me:** to to do today is just show you briefly how we'll work.

[1:11] **Me:** I'll talk a little bit about, you know, what makes rhyme special, but then really focus most of our time on

[1:13] **Me:** you guys and, you

[1:18] **Me:** kind of your interest in voice AI, some of the the challenges you may have come across, which brought

[1:22] **Me:** you to us today. I do have a live demo prepared. I will just use look

[1:23] **Me:** at that very briefly.

[1:28] **Me:** And then, of course, we can just go over any questions, concerns you might have and talk

[1:29] **Me:** about potential next steps. How does that sound?

[1:33] **Voice 2:** Sounds good. Just to frame it,

[1:38] **Voice 2:** we're getting pressure about voice quality, but we do have a lot of fires.

[1:42] **Voice 2:** So more concrete and less shiny stuff, the better.

[1:47] **Me:** Yeah, that sounds pretty good to us.

[1:48] **Me:** So

[1:48] **Voice 2:** I'm just

[1:51] **Me:** Let's see.

[1:51] **Voice 2:** Sorry.

[1:51] **Me:** Just

[1:52] **Me:** go ahead.

[1:57] **Voice 2:** We also have a stack that's already pretty opinionated.

[1:59] **Voice 2:** So wherever this goes,

[2:02] **Voice 2:** integration effort is a big lens for me.

[2:07] **Me:** Yeah. You know, we're we're generally very easily compatible. I I I think

[2:11] **Me:** by virtue of being introduced you with by night. So we are fully compatible with

[2:15] **Me:** Cognizant. So, you know, at the end of the day, it's as simple as a couple toggles and

[2:16] **Me:** in your UI. Obviously, there's a lot more minutiae.

[2:21] **Me:** Involved to picking some of the voices and and doing your security checks

[2:23] **Me:** But, no, it the actual heavy lifting is pretty minimal.

[2:28] **Me:** So today, we're gonna be using a Copilot system just

[2:30] **Me:** because I have terrible handwriting. I hope you don't

[2:35] **Me:** mind. We're we're just gonna be transcribing securely to

[2:35] **Me:** a local encrypted system.

[2:38] **Me:** What you shared today, and then we'll just have an edited summary

[2:43] **Me:** that we can kind of view, together just so that, we make sure everything you you share is

[2:45] **Me:** is really, captured.

[2:52] **Voice 2:** As long as nothing's

[2:54] **Voice 2:** stored or used beyond this, I'm

[2:56] **Voice 2:** cautious about recordings.

[2:59] **Me:** Oh, it's not recorded. It's just like a digital whiteboard.

[3:01] **Me:** Okay.

[3:04] **Me:** Great. Well, I I appreciate your flexibility there.

[3:09] **Me:** So let's see. Now that we've we've covered that,

[3:14] **Me:** me tell you very briefly what makes rhyme special. You know, we've been around since

[3:16] **Me:** 2022, founded by Lilly, Brooke, Ari's.

[3:21] **Me:** Two of our our team members are actually linguists, and then is an engineer. The

[3:22] **Me:** two from Stanford, one from Harvard.

[3:28] **Me:** Data collection is actually where we we start to differentiate

[3:31] **Me:** ourselves. A lot of our competitors use the exact same podcast

[3:35] **Me:** and actor recordings and sort of shared data sets in the industry.

[3:39] **Me:** We actually took a different approach and, started with a a recording studio.

[3:42] **Me:** So we we custom built, everything and, painstakingly

[3:47] **Me:** made recordings of real people, which was a bit of a slower process, but I think

[3:51] **Me:** started to pay off in dividends when you actually hear our voices and realize some of the

[3:56] **Me:** minutiae of language are captured much more eloquently by our approach.

[4:01] **Me:** Than the standard acting one.

[4:01] **Voice 2:** Yeah, we've heard that

[4:02] **Voice 2:** claim from others.

[4:07] **Me:** Well, happy to show you a little later, but you know, we're also

[4:10] **Me:** the proof is in the pudding. Already working with tens of millions of phone calls every month from some of the larger

[4:15] **Me:** restaurant chain, health care providers, telecommunications firms,

[4:15] **Me:** and even governments around the world.

[4:21] **Me:** You know, even recently is

[4:21] **Voice 2:** last

[4:21] **Me:** month we did a study with a

[4:26] **Me:** 100,000 phone calls. And now no one actually loves receiving a

[4:27] **Me:** cold call, so there's lots of hang ups even if a human does it. But

[4:32] **Me:** actually had the lowest hang up rate compared to any of our competitors, as well as the fastest

[4:33] **Me:** call completion

[4:38] **Me:** time. So we're really proud about that. We also, you know, being able to help customers

[4:41] **Me:** significantly increase some of their their sales and performance metrics.

[4:46] **Voice 2:** Okay. But, like, when you say tens of millions of

[4:47] **Voice 2:** calls, are those

[4:52] **Voice 2:** mostly outbound?

[4:52] **Me:** No, mostly in inbound contact centers, just

[4:53] **Me:** like you.

[4:55] **Voice 2:** Okay.

[5:00] **Me:** Yeah. So

[5:03] **Me:** anyway, let let's really dive into you guys. That's that's mostly why we're here today.

[5:06] **Me:** How are you doing speech today? What tools are you using?

[5:11] **Voice 2:** It's a mix. So nice on the court,

[5:13] **Voice 2:** and a couple of other vendors around the edges.

[5:15] **Voice 2:** Whisk quality is okay.

[5:17] **Voice 2:** But

[5:20] **Voice 2:** it can get flat. Customers notice.

[5:25] **Me:** Can you say say more about that? When when you say flat,

[5:29] **Me:** and when you say what customers notice, if you can speak about both of those?

[5:29] **Voice 2:** Sure.

[5:32] **Voice 2:** Just feels robotic at times.

[5:33] **Voice 2:** Off.

[5:37] **Voice 2:** Empathy's not quite right. You get caller saying it sounds

[5:38] **Voice 2:** weird?

[5:42] **Me:** So when callers say that the voice sounds weird,

[5:45] **Me:** if I can ask, I mean, that doesn't sound good, but why does that matter?

[5:50] **Voice 2:** Well, it dense truss

[5:51] **Voice 2:** So in a claims context,

[5:53] **Voice 2:** trust is everything.

[5:56] **Voice 2:** It feels off, they bail or they ask for a human,

[5:59] **Voice 2:** which hits our costs and handle time.

[6:11] **Me:** So cost and handle time. Okay.

[6:15] **Me:** Right. What else matters to you guys?

[6:19] **Voice 2:** Ease of rollout, honestly.

[6:25] **Voice 2:** It's months of lift, that's tough. Also, risk.

[6:25] **Voice 2:** We can't afford downtime or

[6:27] **Voice 2:** escalation spiking.

[6:29] **Me:** How would you measure ease of rollout?

[6:33] **Voice 2:** Short implementation,

[6:35] **Voice 2:** low lift on engineering,

[6:38] **Voice 2:** ideally, over custom.

[6:42] **Voice 2:** And clean handoffs with our existing stack.

[6:46] **Me:** And then in terms of the voice quality issues and

[6:50] **Me:** increasing trust, how would you measure success there?

[6:54] **Voice 2:** Higher containment without hurting CSAT?

[6:59] **Voice 2:** Fewer negative comments about the voice, and

[7:00] **Voice 2:** no increase in handle time

[7:09] **Me:** Monique, what about from your

[7:15] **Voice 2:** If it reduces repeat calls and

[7:17] **Voice 2:** doesn't add a new queue of

[7:20] **Voice 2:** edge cases for my team.

[7:21] **Voice 2:** That's a win.

[7:25] **Me:** Got it. And anything else?

[7:29] **Voice 2:** Just don't create new fires.

[7:33] **Voice 2:** Prove it works in our world, small first.

[7:36] **Me:** In terms of fires, how would you measure that?

[7:41] **Voice 2:** No spike in exceptions, no increase.

[7:45] **Voice 2:** In supervisor escalations. No last minute scrambles.

[7:52] **Me:** Yeah. No. Understood. That makes sense. And now, Raj, you you own the tech stack.

[7:54] **Me:** Any additional problems or success criteria?

[7:58] **Voice 2:** Reliability and clean integration.

[8:01] **Voice 2:** If it plays nice with nice,

[8:05] **Voice 2:** pardon the pun, and add fragility,

[8:06] **Voice 2:** Cool.

[8:13] **Me:** And what provider do you use at the moment?

[8:17] **Voice 2:** Shaken.

[8:23] **Voice 2:** Raj says they are using what's bundled with their

[8:24] **Voice 2:** current platform plus one other TTS vendor.

[8:26] **Voice 2:** For specific flows. But

[8:27] **Voice 2:** didn't name them.

[8:32] **Voice 2:** Alina adds that they've experimented with a few,

[8:35] **Voice 2:** and hit the same quality salient complex conversations.

[8:38] **Voice 2:** Monique isn't shopping for change just for change's sake, but

[8:41] **Voice 2:** she'd need to see a measured improvement.

[8:42] **Me:** So Now

[8:42] **Voice 2:** she's asking

[8:47] **Me:** So far, I've heard about, the voice quality issues and those affect

[8:49] **Me:** trust which increase costs and handle time.

[8:51] **Me:** And then

[8:56] **Me:** your your stack is already opinionated, so the the impact

[8:58] **Me:** that is sort of the the risk of system stability.

[9:02] **Me:** And then, you you rollout requiring months of effort would also be a problem.

[9:05] **Voice 2:** That's pretty fair. And

[9:08] **Voice 2:** yeah, we don't have Fimp in its cycles.

[9:12] **Me:** Totally get that.

[9:19] **Me:** Anything else

[9:21] **Me:** that we haven't covered here or that we should think more deeply about?

[9:25] **Voice 2:** For me, risk and rollout.

[9:30] **Voice 2:** Prove it small, show the metrics, then maybe.

[9:33] **Voice 2:** If you can meet us where we are,

[9:35] **Voice 2:** then we can keep talking.

[9:37] **Me:** That makes a lot of sense.

[9:40] **Me:** Great. Well,

[9:47] **Me:** let me actually just jump into a quick demo. ChatGPT

[9:49] **Me:** don't interrupt again until I, say the demo is over.

[9:54] **Voice 2:** Got it. Staying quiet until you say the demo is over.

[9:57] **Voice 2:** Hi there. Tell me today.

[10:00] **Voice 3:** Could you please provide your five digit ZIP code for verification?

[10:03] **Me:** Hey, Soria. Could you start from the top?

[10:08] **Voice 3:** Of course. If you have a damaged device, you'd like

[10:09] **Voice 3:** to claim, I can guide you through the process.

[10:13] **Voice 3:** Could you please provide your five digit ZIP code first for verification?

[10:14] **Me:** 94102.

[10:21] **Voice 3:** Thanks, Colin. Your policy is active and covers accidental damage.

[10:23] **Voice 3:** When did the damage to your iPhone 16 Pro happen?

[10:28] **Me:** July

[10:28] **Voice 3:** Was the damage accidental?

[10:33] **Me:** Yes. I was just getting ready for this really important presentation. I was so excited.

[10:37] **Me:** I accidentally, knocked my phone off the table and it cracked the screen.

[10:42] **Voice 3:** Does your iPhone still tower on at work despite the cracks?

[10:42] **Voice 3:** Screen?

[10:43] **Voice 4:** Yep.

[10:46] **Voice 3:** Do you still have your phone with you?

[10:47] **Voice 4:** I

[10:47] **Me:** do.

[10:56] **Voice 3:** To confirm, on July 11, you accidentally dropped your iPhone 16 Pro cracking the screen, but it still powers on and you have the device with you.

[10:58] **Voice 3:** Is that correct? Yes.

[11:05] **Voice 3:** Your claim numbered CLM demo,

[11:07] **Voice 3:** 2049 is submitted.

[11:10] **Voice 3:** The deductible for the screen repair is $29.

[11:14] **Voice 3:** I can send a secure document upload link to your phone to complete your claim.

[11:16] **Voice 3:** Would you like me to send it now? Yes.

[11:27] **Voice 3:** The secure upload link has been sent to your phone ending in 0148. Would you like to finish up here or speak with a device claims specialist for any further help?

[11:29] **Me:** Specialist would be great, thanks.

[11:35] **Voice 3:** I'm transferring you to a device claim specialist now.

[11:36] **Voice 3:** Thank you. They have all your

[11:36] **Voice 4:** detail.

[11:41] **Voice 3:** Go ahead. So you need to repeat anything.

[11:43] **Voice 3:** Please hold

[11:43] **Voice 2:** that on. Load well.

[11:46] **Voice 2:** Okay. Stopping until you say otherwise.

[11:48] **Me:** Yeah. Go ahead with the transfer.

[11:51] **Voice 2:** It sounded smooth and natural.

[11:54] **Voice 2:** We wanted to sharpen it, I

[11:55] **Voice 2:** okay.

[11:56] **Me:** Pausing.

[11:59] **Me:** Where is the specialist

[11:59] **Voice 4:** available?

[12:07] **Voice 5:** Colin, I'll prepare your information for the specialist so you won't have to repeat anything.

[12:08] **Voice 5:** One moment, please.

[12:13] **Voice 5:** Colin, I'm ready to transfer you now. Your information about the

[12:17] **Voice 5:** accidental damage claim for your iPhone 16 Pro on July 11 is with the specialist.

[12:19] **Voice 5:** Please hold for a moment.

[12:21] **Voice 5:** Connecting you to the device claims specialist.

[12:24] **Voice 5:** Your summary travels with you, so nothing needs repeating.

[12:42] **Me:** Oui, merci beaucoup, c'est gentil, votre assistance. Je

[12:43] **Me:** juste confirmer, tout est bon

[12:59] **Voice 3:** Si vous avez besoin d'autre chose, je suis là pour vous aider.

[13:02] **Voice 4:** Non sais-tu.

[13:05] **Me:** So what did you think of that call?

[13:10] **Voice 2:** It felt smooth and human the pacing was

[13:11] **Voice 2:** steady.

[13:14] **Voice 2:** The confirmation step was clear, which I think

[13:16] **Voice 2:** builds trust.

[13:21] **Voice 2:** If we were tightening it even more, maybe trim up

[13:21] **Voice 2:** couple of the confirmation lines, but

[13:23] **Voice 2:** honestly, that was pretty solid.

[13:26] **Me:** I really appreciate that feedback.

[13:29] **Me:** Large, anything to add?

[13:33] **Voice 2:** Just make my life harder, but

[13:36] **Voice 2:** the quality did sound better than what we're used to.

[13:42] **Me:** Yeah, I mean, you know, I picked this pretty specially. We,

[13:45] **Me:** started with thousands of voices when we were training our latest model, Coda, but we

[13:49] **Me:** went through a very rigorous process of pairwise comparison to

[13:54] **Me:** battle the voices off and ended up with about a 100 that you can

[13:59] **Me:** select from. And some of our customers do that based on region or

[14:00] **Me:** even, demographics of the the caller.

[14:05] **Me:** I picked Vespera for the primary voice.

[14:08] **Me:** Obviously, there are some others there as well because of her sort of formal professional nature.

[14:11] **Me:** And of course, she's American.

[14:15] **Me:** What for these qualities would feel right for your brand?

[14:20] **Voice 2:** Warm but competent. Not overly cheer

[14:22] **Voice 2:** cheerful, not call center y, just

[14:27] **Voice 2:** reassuring. Like someone who knows what they're doing.

[14:31] **Me:** Could you walk me through your decision process if we did have a successful pilot?

[14:34] **Voice 2:** It depends, but

[14:37] **Voice 2:** operations, IT, CX,

[14:38] **Voice 2:** and then procurement.

[14:40] **Voice 2:** Consensus driven.

[14:45] **Voice 2:** Risk first.

[14:45] **Me:** And, what teams besides those here today would usually need to be involved?

[14:50] **Voice 2:** IT security for sure.

[14:51] **Voice 2:** Maybe legal depending on

[14:52] **Voice 2:** data handling.

[14:57] **Me:** Are there any procurements or security steps we should plan for early?

[15:00] **Voice 4:** Check.

[15:05] **Voice 2:** Raj would want a security review upfront.

[15:08] **Voice 2:** Documentation on data handling and encryption.

[15:12] **Voice 2:** Monique wants a clear pilot plan with success metrics,

[15:16] **Voice 2:** Procurement isn't the blocker if the business case is

[15:16] **Voice 4:** tight.

[15:21] **Voice 2:** But security could take a few weeks. So get ahead of it.

[15:24] **Voice 2:** And expect reference checks with similar customers.

[15:29] **Me:** Yeah. No. That makes perfect sense. And, what would be a realistic time frame

[15:29] **Me:** on your part?

[15:32] **Me:** One moment.

[15:38] **Voice 2:** Raj estimates four to eight weeks for a contained pilot if

[15:48] **Voice 2:** security review proceeds smoothly. Monique wants narrow scope first not a big rollout with a data driven gono go after that.

[15:52] **Me:** And, what who would also need to sign off on a project like this?

[15:56] **Voice 2:** Checking ones what?

[15:59] **Voice 2:** Monique says it's collective, basically.

[16:10] **Voice 2:** Raj for the technical side and security, Elena for CX impact, Monique for operations, then procurement and an executive sponsor for final sign off.

[16:15] **Voice 2:** And if all the stakeholders feel their concerns are met,

[16:17] **Voice 2:** with measurable results and

[16:19] **Voice 2:** low risk, that's

[16:19] **Voice 4:** what

[16:19] **Voice 2:** gets him in the line.

[16:24] **Me:** Got it. She said the biggest problem was voice quality still, though.

[16:27] **Voice 2:** Yeah. It's hurting trust.

[16:36] **Me:** Understood.

[16:43] **Me:** Great. So what questions can I answer for you guys?

[16:48] **Voice 4:** Probably just

[16:51] **Voice 2:** send a short summary of what success would look like in a pilot

[16:53] **Voice 2:** Keep it high level.

[16:55] **Voice 2:** We'll regroup internally from there.

[16:59] **Me:** And what else can I tell you?

[17:01] **Voice 2:** That's it for now.

[17:04] **Me:** Alright. Well, let me recap what I've heard here today.

[17:09] **Me:** Sounds like some of the biggest issues are voice quality and, risk and rollout are really important.

[17:12] **Me:** And compatibility with your existing stack.

[17:14] **Me:** And

[17:19] **Me:** yes, success is measured by reliability and needing to see

[17:21] **Me:** measured improvement before considering change.

[17:22] **Me:** Am I missing anything?

[17:26] **Voice 2:** Now that's accurate.

[17:32] **Me:** Well, let me just talk you briefly through what happens next. Usually, we'd kick off

[17:34] **Me:** and to align on a scope and success criteria, then

[17:39] **Me:** pilot on a live call flow and limited production, then reconvene to review the results,

[17:40] **Me:** and then measure game success. You know?

[17:45] **Me:** CSAT scores, call completion, for example, or caller's sentiment.

[17:46] **Me:** Then we can convene afterwards.

[17:48] **Voice 4:** Okay.

[17:53] **Me:** Okay. Well, thank you so much.

[17:55] **Me:** I think that's everything from my end.

[18:00] **Me:** Let's try a follow-up next Tuesday.

[18:04] **Voice 2:** Let me check.

[18:07] **Voice 2:** Yeah. That works. Send the one pager before then?

[18:09] **Voice 2:** And we'll take a look.

[18:10] **Me:** Thanks guys.

[18:14] **Voice 2:** Thank you. We appreciate the conversation.