Every time you leave a voicemail, talk to someone in a call center or join a conference call, data is being created. Tons of data. While voice assistants like Alexa and Google Home get most of the attention, millions of hours of voice data sit idle in today’s enterprises, leaving a tremendous amount of potential value on the table.

If voice data is an untapped resource, what’s stopping companies from leveraging its full value? Voice is far more difficult to secure, deliver and analyze than traditional data. It presents unique challenges for machine learning applications that require clean, diverse, representative data. But it’s also an exciting time in the voice computing space, and companies that can overcome these challenges will reap the benefits of this new frontier.

Enterprise Voice Data Is Everywhere

While traditional chatbots may have passed their zenith, conversational AI — the ability to interact with computers via speech — is unequivocally on the rise. In fact, 1 in 5 searches on Google’s mobile app are voice searches, and Amazon Alexa recently surpassed 10,000 skills. (Full disclosure: Amazon and IBM are Delphix partners.) Speech interfaces will continue to expand into Internet of Things (IoT)-enabled platforms like cars, smartphones and internet browsers. And they’ll be able to hold longer, more intelligent conversations. A study from Juniper Research forecasts that advanced chatbots will save the customer service industry $8 billion a year by 2022.

But it’s not just about speech interfaces. There are millions of existing voice interactions with humans every day such as customer service calls, health care provider interactions and more. Companies like Pindrop are augmenting these calls with AI to detect fraud, but we’re only starting to scratch the surface. Imagine mining existing call center data to calculate the likelihood that customers will purchase your product or deliver real-time customer satisfaction metrics.

Companies need to stop thinking about voice data as just an interface or the outcome of a conversation. Rather, voice data is a unique source of critical insights for the business.