Rime Raises $24 Million to Push Enterprise Voice AI Beyond IVR

Updated July 15, 2026 4:54 pm

Rime, a San Francisco voice AI startup focused on enterprise phone calls, has raised $24 million in Series A funding as it tries to stand out in a crowded market by building its own conversational training data and lower-latency speech-to-speech models. The round, announced Wednesday, was led by M13 Ventures and underscores investor interest in companies trying to make AI phone agents reliable enough for regulated business use.

The company says its models are designed to improve how businesses handle customer calls in areas such as sales, support, healthcare, aviation, and financial services. Rather than depending on scraped audio from the internet, Rime records its own conversations in a San Francisco studio and uses that data to train models that better understand brand names, industry jargon, and pronunciation edge cases.

Why Rime thinks enterprise voice AI still has a long way to go

Rime’s pitch is that the voice AI market has matured quickly on the surface, but the underlying experience still leaves room for improvement. Enterprises are already moving calls to AI vendors for customer service, marketing, and sales, yet many systems still feel like upgraded versions of old interactive voice response, or IVR, menus rather than truly natural assistants.

Rime co-founder and chief executive Lily Clifford argues that the industry has made it easier to assemble voice products, but not necessarily to create a better customer experience. In her view, the basic problem is not just getting a system to speak fluently; it is making it respond quickly, handle interruptions smoothly, and sound reliable in high-stakes business settings.

Clifford said voice AI remains short of what enterprises need to automate most phone interactions, adding that today’s tools can improve the sound of a call without necessarily making the experience feel dramatically better for the person on the other end.

That challenge is central to the competitive race in enterprise voice AI. Companies such as ElevenLabs and Deepgram are supplying core voice technology, while infrastructure providers including Vapi, Retell, and LiveKit are helping developers build call workflows. At the application layer, dedicated customer support vendors like Sierra and Decagon are trying to own the end-user experience.

How Rime is trying to differentiate itself

Rime is betting that owning the data pipeline will create a durable technical advantage. Instead of treating training data as something to be gathered from the broader web, the startup has built a studio to record conversations directly, giving it more control over the kinds of speech patterns, terminology, and conversational timing its models learn.

The company says that approach makes its models better at handling the names of brands, products, and industry-specific terms that often trip up general-purpose systems. It also aims to reduce the amount of customer customization required when a company wants to deploy voice AI in a specialized vertical such as healthcare or aviation.

What is a phoneme-based architecture?

A phoneme-based architecture is a system that focuses on the smallest units of sound in language rather than relying only on broader text patterns. Rime says that design helps its models adapt more precisely to different pronunciations, which can matter when a call center agent must recognize uncommon terms, local names, or technical vocabulary under time pressure.

That technical emphasis matters because enterprise buyers often care less about novelty than about consistency. If a voice agent repeatedly mispronounces a medication, a destination, or a financial product, the system can lose trust quickly and become a liability rather than a cost-saving tool.

What the $24 million Series A means for Rime

The financing gives Rime more room to scale both its research and commercial operations. The company said it plans to grow its workforce of 35 employees, with hiring targeted at model development, engineering, and partnerships. The new capital also deepens support from investors who believe the market has not yet sorted out which layer of the voice AI stack will capture the most value.

M13 Ventures led the round, and Twilio Ventures, Corazon Capital, Unusual Ventures, and other existing backers also participated. M13 partner Morgan Blumberg is joining Rime’s board as part of the transaction.

The company had previously raised $5.5 million in seed funding in May of last year, suggesting the new round arrives after a relatively quick validation cycle for a young startup still building out its technical roadmap.

Why investors are still backing new voice AI infrastructure

Enterprise voice AI has become one of the most active corners of the broader artificial intelligence market because phone calls remain a huge, expensive operational channel. Even as chatbots and web-based support tools have become more common, large organizations still receive a significant share of customer requests by phone.

That makes the sector attractive for startups that can reduce wait times, cut staffing costs, or route calls more efficiently. It also explains why the ecosystem has developed so quickly across multiple layers: foundational model providers, orchestration tools, developer platforms, and packaged customer service apps.

Rime’s opportunity lies in convincing buyers that reliability and voice quality are not enough on their own. The startup believes the bigger differentiator is whether an AI system can perform well across real-world enterprise scenarios without requiring a long and expensive customization project.

How crowded is the market?

The market is crowded enough that nearly every layer now has serious competition. Model companies are offering higher-quality speech generation, infrastructure firms are simplifying deployment, and application startups are packaging voice AI into turnkey support products. That pressure is pushing startups to specialize, whether by industry, latency, or the quality of underlying training data.

• Model layer: Voice generation and speech understanding
• Infrastructure layer: Call routing, orchestration, and deployment
• Application layer: Customer service and sales workflows

Rime is choosing to compete primarily on the model layer, while still claiming enterprise usefulness at the application level through better performance on specialized language and pronunciation.

How Rime’s product strategy is changing

The startup originally built its system as a combination of separate speech-to-text, text-to-speech, and large language model components. That architecture is common in early voice AI products because it is relatively straightforward to assemble, but it can also add complexity and delay as audio moves through multiple stages.

Rime is now shifting toward speech-to-speech models, which are intended to make conversations more fluid and faster. The company says this change should lower latency, improve how the system handles turn-taking, and reduce errors when people speak over one another or when background noise is present.

Another goal is to decrease the amount of orchestration needed behind the scenes. In practical terms, that means Rime wants fewer moving parts between the caller and the AI response, which could simplify operations and make the product easier to maintain at scale.

Why latency matters in enterprise calls

Latency matters because even small pauses can make a voice agent feel robotic or frustrating. In a support call, delays can cause people to repeat themselves, interrupt the system, or abandon the interaction entirely. For businesses, that can undermine efficiency and customer satisfaction at the same time.

Speech-to-speech systems are attractive because they aim to compress the whole interaction pipeline. If successful, that could create a more natural conversation and help AI agents feel less like a scripted menu and more like a responsive assistant.

Which customers is Rime targeting?

Rime says its customers span food service, healthcare, airlines, and fintech. Those sectors all depend heavily on voice interactions and are often under pressure to balance service quality with labor costs and compliance demands.

The company says clients include Mayo Clinic, Dialpad, Upstart, and Asurion. It claims that its data and model approach help keep callers engaged longer, which it presents as evidence that the technology can support more successful enterprise contracts and more usable conversational experiences.

That is an important claim in a market where buyer confidence depends on practical outcomes. Enterprises are unlikely to adopt voice AI widely unless it can retain user attention, resolve issues efficiently, and perform consistently enough to satisfy both customers and regulators.

Who is leading the technical push at Rime?

Rime recently hired Rafael Valle as chief scientist, adding an executive with deep experience in audio understanding research. Valle previously worked at Meta Superintelligence Labs and on NVIDIA’s applied deep learning audio research team, which gives Rime another signal to market that its technical ambitions are serious.

For a startup competing against better-known names and broader platforms, personnel can matter as much as fundraising. Senior talent helps shape model strategy, improve performance benchmarks, and communicate credibility to enterprise buyers who may be evaluating several vendors at once.

Blumberg of M13 said the market is still far from settled and argued that Rime’s focus on low latency, reliability, and regulated environments makes it notable among voice AI startups chasing enterprise demand.

What does this say about the future of voice AI?

This funding round suggests the next phase of voice AI will be less about proving that AI can answer calls and more about proving that it can do so well enough for demanding business use. The bar is rising from novelty to operational dependability, especially in sectors where mistakes can affect revenue, compliance, or patient experience.

That shift could favor companies with stronger data pipelines, more focused technical architectures, and products built for specific industries rather than generic voice demos. It could also push the market toward consolidation as enterprises choose vendors that can combine model quality with dependable orchestration and customer support workflows.

For Rime, the challenge will be to convert its technical thesis into sustained enterprise traction. The new funding provides capital and credibility, but the company still has to prove that its data strategy and speech-to-speech roadmap can deliver a meaningful advantage over better-capitalized rivals.

Timeline of Rime’s growth

The bigger enterprise voice AI race

Rime’s latest round is one more sign that voice AI remains one of the most aggressively funded areas in applied artificial intelligence. Companies are not just racing to create more human-sounding voices; they are trying to own the systems that can safely and consistently handle business-critical interactions.

That race is unfolding across several competing visions. Some companies are building the core models, others are building the plumbing, and others are trying to deliver full-service customer-facing agents. Rime is positioning itself somewhere between a model company and a vertical enterprise vendor, with a message built around data quality, pronunciation accuracy, and low-latency performance.

Whether that is enough to become a category leader will depend on execution. For now, the $24 million Series A gives Rime the resources to keep refining the technology, hire more talent, and make its case that the next leap in enterprise voice AI will come from better models, not just better wrappers around them.