Automated cloud backup provider Druva Inc. said today that it’s incorporating generative artificial intelligence into its platform to simplify customer interactions.

The copilot, called Dru, uses a conversational interface (pictured) to help users more quickly pinpoint the cause of a problem and identify a solution. It lets users request real-time, custom reports, ask follow-up questions and act upon suggestions provided by the chatbot based on Druva documentation and troubleshooting information. The software can analyze logs, troubleshoot errors for quicker alerts and provide explanations that make it easier for users to analyze and find relevant information.

Dru also advises on how to leverage new and advanced functions to get more value from the platform and can help with everyday administrative tasks such as triggering new backups of specific workloads.

Just ask

“There’s not a company in the world that wakes up and says, ‘I really want to spend a lot of time on my backup application today,’” said Stephen Manley, Druva’s chief technology officer. “But there are times when something will go wrong. If someone hasn’t been on a product for an extended period of time, they may not instinctively know how to use the interface. They want to just ask a question.”

Manley said that most of the information Dru provides is already available through menus and queries in the Druva platform. Generative AI is essentially a replacement for existing navigation schemes that provides a faster track to diagnosis and resolution.

“We have over 5,000 customers, so we’ve got a fairly strong history of understanding what sort of things go wrong, what those signatures look like and how best to resolve them,” he said. “We’re using that historical knowledge to guide users because, while an issue may not be new to us, it might be the first time it has happened to them.”

Druva didn’t choose a single training model for Dru but used Amazon Web Services Inc.’s Bedrock managed service to experiment with a variety of foundation models.

“It lets us swap out different models as we continue to test and evolve and because it’s inside of Bedrock, nothing on top needs to know that the underlying model has changed,” he said. “It’s too early to make an all-out bet on one horse.”

Manley said that the generative AI features are in early access availability and should be rolled out over the next quarter. They will become part of the core product at no additional charge.

Image: Druva