A revolution is underway in customer communication, guided by voice rather than keyboards or screens.
According to Andreessen Horowitz of Venture Capital Firm, voice-based artificial intelligence (AI) agents have been a promise longer than products, and are now pushing ahead to the point that they are beginning to replace human labor in industrial areas from healthcare to retail.
“Voice is one of the most powerful unlocks for AI applications companies.” Olivia MooreAndreessen Horowitz partner wrote in the 2025 AI Voice Agent Update Blog post. “This is the most frequent and information-dense communication, and it's the first time it's been programmable for AI.”
Creating something audio programmable means that AI can interpret and respond and Act on voiced queries that are more accurate and reliable. The voice is not naturally structured and messy. People suspend topics, change topics, and use slang.
Moore said voice AI Let's forgive Businesses are open 24/7 for customers Instead of You will have to wait until you are placed in the office. For consumers, “We believe that this is the first and perhaps the main way people interact with AI.”
According to the PYMNTS Intelligence Report, 30.4% of Gen Z consumers are already shopping by voice every week. In “Global Digital Ways,” the report showed that millennials ranked second with 27.6%. At all ages, the average is 17.9% of consumers.
Last year, Voice AI startup It was raised According to research firm CB Insights, $2.1 billion has risen eight times since 2023. The 2024 bumper crop included ElevenLabs' $180 million funding round.
According to the research company, growth was driven by advancements such as Openai's real-time API for speech-to-speech applications, such as Openai's real-time API for speech-to-speech applications.
“It's been the last 12-18 months that I've seen AI voice agents perform just like humans,” says Alex Levin, co-founder and CEO of Voice AI Company Regal. I said Wall Street Journal.
March, Yam! Taco Bell, a brand owned by KFC and Pizza Hut, announcement partnership It uses NVIDIA, which deploys AI solutions. This includes the implementation of voice AI across call center brands to process phone orders When demand rises rapidly.
I have a jersey microphone Tap Soundhound's AI, which places audio orders at 50 stores. Recently, Soundhound Partnership Expand “Allina Health will deploy “Alli,” an AI agent that can answer calls from patients. Helps patients manage appointments, quickly refill medications, find doctors and locations and Answer non-clinical questions.
read more: How the world is digital: delve deeper into global digital engagement
A crucial moment in audio ai
in Over the past year, the AI infrastructure that underlies voice has been It has been fundamentally improved.
One year ago, Openai announcement “Audio Mode” built on top GPT-4O It provided real-time voice responsiveness, ability to disrupt, and versatility in It's an emotional tone, not a robot's response.
In November, conversational AI continued in November, and last month version 2.0 was announced. Meanwhile, players like Kyushu and Speechmatics have brought real-time full-duplex conversations to the production, Moore said.
These models have become more affordable as latency decreases. According to Moore, Openai reduced the cost of the GPT-4o API by up to 87.5% last December.
“Conversation quality is now a largely resolved issue,” Moore said, and startups are competing to deploy voice as “wedges” or as entry points to broader enterprise platforms, as their business is small.
Ketan Babaria, Chief Digital Head of Insurance Marketplace Ehealth, told the Journal that Voice AI is incredibly good.
“All of a sudden I realized these agents were becoming very human,” Bavaria said. “We've reached a point where our customers can't distinguish between the two.”
The next advance is AI voice agents that can use mobile phones do Independent tasks such as restaurant reservations, sales deadlines, etc. and When they ordered, Polyai CEO Nikola Mrksic told the Journal.
According to Moore, voice AI use cases include:
- Out of business hours or overflow calls: These are calls that otherwise go to voicemail. Voice agents can collect or share information, and even complete reservations and transactions.
- Net New Outbound Calls such as Customer Check Calls, Activation Calls, Lead Calls and Similarly.
- The back office calls vendors, suppliers and more.
Despite rapid adoption, Voice AI is still facing hurdles. The risk of reputation is high, especially if the system is not publicly available. For example, McDonald's That voice pulled AI pilot Using IBM after video of failed orders went viral.
read more: Humanity begins to add voice modes to AI Assistant Claude
read more: Openai will debut advanced voice AI for subscribers
read more: T-Mobile Parents develop voice-activated AI phones
