Exclusive: Deepgram Launches Free Enterprise Voice AI Platform to Entice Potential Clients
Enterprise speech technology startup Deepgram is offering a limited version of its platform for free. The new MissionControl setup lets companies try out Deepgram’s automatic speech recognition training system as a way of enticing them to sign up for for a more complete package of Deepgram’s technology.
Deepgram’s platform offers a way for businesses to record, transcribe, and analyze audio ranging from customer service calls to internal meetings. Where Deepgram claims to differ is that it can improve the accuracy of automatic speech recognition without slowing down or using more resources because it uses end-to-end deep learning. Deepgram can vary the precision, and the price of its platform, depending on a client’s needs, but even at the basic level, Deepgram is supposed to be more accurate than is usually the standard for voice AI. From that point, each speech model is customized through training to function better in its environment, learning industry jargon, and getting better at filtering out irrelevant background noise.
MissionControl brings all of the data labeling and model training of Deepgram’s usual setup but is aimed at attracting more potential customers because it is free. The engineers, data scientists, or product developers at a company looking to incorporate this kind of voice technology can test it out with their own custom speech model. MissionControl’s free platform includes 20 hours a month of automatic speech recognition supported by ten minutes of professional data labeling, two training-ready datasets, two custom models, and a cloud deployment. Once they’ve had a taste of the process and its functions, they may decide they need more than is available for free and talk to Deepgram about integrating the AI into their company’s tech.
“It’s something you can’t get elsewhere, turning a knob to make it more accurate,” Deepgram CEO Scott Stephenson told Voicebot in an interview. “When we started, this approach was new, so it was handled internally, and we were increasing accuracy in a way not apparent to the user. Now we are allowing people to [setup and train] models themselves.
Deepgram had been thinking about creating a free version of MissionControl for a while. Still, it took the $12 million the company raised in March to give the company enough resources to devote to the project. That funding also contributed to the $1 million worth of Deepgram’s platform that the company donated to assist medical providers as the current COVID-19 health crisis has grown. Despite the name sounding like NASA headquarters, MissionControl is unrelated to the further strategic investment by U.S. intelligence community-founded venture capital firm In-Q-Tel that followed earlier this summer. The biggest reason that now was a good time for the free platform to come out is as much awareness and interest as anything else, according to Stephenson.
“If you tried to release this two years ago, people wouldn’t have understood what it was for or what it could do for them,” Stephenson said. “Now, they’re actually becoming more data-driven; they’re not just saying it anymore. Now, if you don’t have data, there’s something wrong.”
The increased awareness of the value of voice AI for enterprise services translates to more competition as well. Some fellow startups, like Otter and Fireflies, are picking up millions in investment each, while others are fodder for lucrative acquisitions, as happened when Cisco bought Voicea. And of course, the same pillars of the tech industry that made MissionControl as a free platform viable are also looking to take up as much of the enterprise side of things as possible.
“It’s a double-edged sword because we also compete with the big tech companies,” Stephenson said. “But, the way people become aware of voice and how it is useful for enterprise is through consumer voice assistants. That’s okay, because they may try Amazon or Google, but trying us they’ll see what we have is unique. Ours is just more accurate.”