Voice turns one
We launched voice in October 2024, and Sierra now handles more phone calls than chats. Thanks to voice, our growth trajectory is looking more like a helicopter taking off than a hockey stick. This year alone, just 17 months since launching publicly, our platform will power hundreds of millions of conversations for hundreds of companies.
In a series of articles this week, we’ll explore how we climbed the voice hill — from load testing at scale, to multilingual support, to outbound calling, optimizing latency, and fine-tuning models. But first, here’s why we believe it’s proven to be such a game changer.
Voice is the most ergonomic, lowest-friction form of communication
Consumer surveys show that, given a choice, customers would rather talk to companies when they have an issue, especially in high-stakes industries like financial services and healthcare. But until now, talk was expensive: usually between $10 and $20 per call. Lowering that cost from dollars to cents enables more companies to talk to more customers in more meaningful ways.
Voice has become programmable infrastructure
AI has digitized the last major analog communication channel: the public switched telephone network.
That shift has turned conversations from one-and-done interactions into data that can be logged, analyzed, and improved. Calls can be A/B tested like landing pages, voice flows refined like software, and performance benchmarked across millions of interactions. Companies can now measure, repeat, and optimize what once required immense cost and complexity to scale.
Voice and chat work as one
Chat will always matter, and many companies start there because it’s familiar territory. With Sierra’s build once, deploy everywhere approach, the underlying logic, data, and integrations stay the same — only the surface changes. When your chat agent learns a new policy, product, or language, your voice agent does too, and vice versa. The result is consistency at scale: customers get the same clarity whether they type or talk, and teams have to improve and maintain one system, not many.
Voice simulations give brands the confidence to go live quickly
Real customer conversations are messy, full of interruptions, background noise, jargon, numbers, accents, and emotion. And when you’re speaking, milliseconds matter. Voice Sims help ensure your agent can filter signal from noise, get the tone right, and listen, think, and speak all at the same time before they talk to a single customer — which is key if you want to launch in weeks, not months.
The conversation will become the interface
As we look to the future, it’s increasingly clear that voice isn’t just a technical shift — it’s a complete rethinking of how we interact with computers. In almost every sci-fi movie, whatever the decade, people talk to machines. Until now, that’s been fiction. We’ve always needed intermediaries: punch cards, command lines, graphical user interfaces, browsers, touchscreens. But Hollywood had it right all along: the conversation is the interface, and that future is a lot closer than you might imagine.
