Dialpad Unveils DialpadGPT, Its Own ChatGPT For The Enterprise

“We’re building our own ChatGPT system that’s going to help us do customized AI for all of our customer intelligence and collaboration needs. The reason why we’re doing that is because it lets us gain more control over it, it lets us scale, it lets us do customizations. It’s also in real-time, which is another big deal. If you look at ChatGPT, it can be over 10 seconds for a response. That doesn’t work obviously, with customer communications in real time,” Dialpad co-founder Brian Peterson tells CRN.

Cloud communications provider Dialpad has harnessed its strengths in AI-powered customer intelligence with the creation of DialpadGPT, the company revealed on Tuesday.

DialpadGPT is a large language model (LLM) that the vendor said has been built on more than five years and five billion minutes of proprietary conversational data. Unlike existing LLM models such as ChatGPT, Dialpad’s generative AI offering caters specifically to business language by pulling in conversational and messaging data across customer service, sales, recruiting, and employee collaboration, according to the company.

For Dialpad, honing real-time AI has been a work in progress for the last five years, Brian Peterson, CTO and co-founder of Dialpad, told CRN.

“We’re building our own ChatGPT system that’s going to help us do customized AI for all of our customer intelligence and collaboration needs. The reason why we’re doing that is because it lets us gain more control over it, it lets us scale, it lets us do customizations. It’s also in real-time, which is another big deal. If you look at ChatGPT, it can be over 10 seconds for a response. That doesn’t work obviously, with customer communications in real-time,” he said.

[Related: Dialpad Launches Partner Portal On Heels Of ‘Record-Breaking’ Growth: Exclusive]

A self-proclaimed AI-based communications leader, Dialpad has already been doing AI with real-time transcription via its own platform, which has made it much easier to develop DialpadGPT from its own customer data, Peterson said.

Once available, DialpadGPT will help enterprises automate tasks, including summarizing sales, customer service, and team engagements and follow-up actions, according to the San Francisco-based company.

“[DialpadGPT] is unique because it’s all our own proprietary AI. A lot of other vendors mentioning AI in the collaboration and communications market are leveraging a third party, which means that cost is added for them. There’s [also] a lack of customization and inaccuracy when using a third-party [and] it’s really hard to do real-time if you don’t have it baked into your platform.,” he said. “All of our collaboration aspects, the calling, meetings, video, even the AI, we build everything ourselves. That is our biggest strategy advantage.”

Dialpad has more than 50 dedicated data scientists, machine learning engineers and linguists that help to customize the platform for specific business needs. Features that Dialpad’s platform offers today includes real-time sentiment analysis for problems that may come up during a live call and real-time coaching for agents that may be speaking too fast.

Customization is critical for businesses that want to tap into LLMs and it’s a big differentiator for Dialpad, Peterson said. “When we are customizing things for specific sales and customer support calls, it makes it more accurate if you’re not using AI based on the entire world wide web of information which isn’t relevant to the business conversations.”

For solution providers, the use cases will be easy to show and sell, Peterson said. “[Partners] can actually try it out in real-time with their customers. There’s a lot of AI being announced that isn’t really available and it’s not available because no one can scale it. And no one can steal it because they’re using a third-party and the third party doesn’t have resources, or it’s too expensive. [But] we’re trying to take it all into our own hands,” he said.

DialpadGPT will be generally available by October 2023.

San Francisco-based Dialpad has a deep bench of collaboration talent. Dialpad’s co-founder and CEO Craig Walker previously served as founder and CEO of GrandCentral Communications, which became Google Voice. Peterson, for his part, was co-founder and vice president of product for UberConference before his time at Dialpad and before that, served as a senior software engineer for Google where he worked with Walker on the Google Voice product.