Cisco powers secure, responsible artificial intelligence innovation at scale with MongoDB

THE CHALLENGE

Using MongoDB solutions to support gen AI for Cisco

As companies rush to adopt generative AI (gen AI), many may be introducing security risks by not following best practices — like using encryption in databases. Cisco, a global provider of networking, collaboration, and cybersecurity solutions, seeks to balance innovation with security as it rolls out hundreds of gen AI applications.

Cisco turned to MongoDB, a long-standing partner of the company, to streamline the path from experimentation to production while maintaining strict security standards. Using MongoDB’s expertise and technology, Cisco aimed to create a secure data environment that would help its developers innovate faster with gen AI.

THE SOLUTION

Providing a unified, secure vector database to build gen AI apps

MongoDB achieved Cisco’s rigorous security certifications in record time, earning approval to handle the company’s most sensitive data, including source code. Once the certifications were in place, Cisco’s Security and Trust Organization worked with MongoDB to create sandbox clusters using MongoDB Atlas. The team also employed MongoDB Atlas Vector Search to power Cisco’s retrieval-augmented generation (RAG) capabilities. Teams can safely experiment with gen AI in these environments without risking data leaks or compromising security best practices.

“Developers are free to focus on their tasks from the beginning without having to check boxes for approval or worry about encryption or other security measures,” says Omar Santos, distinguished engineer at Cisco. “MongoDB helps us accelerate the development process.”

When Cisco was assured that its internal data was secure, the company began to expand its use of gen AI. The company used MongoDB Atlas Vector Search to enhance and further expand the impact of its WebEx virtual assistant. “The virtual assistant has been in production for over three years and has successfully deflected or resolved more than 100 support cases,” says Mustafa Kadioglu, lead data scientist at Cisco. “This has saved us roughly $10 million.”

The assistant uses an RAG architecture to semantically index WebEx’s 3,600 help articles and dynamically generate responses to user queries. A caching layer built on top of MongoDB Atlas Vector Search maintains low latency as usage grows.

THE RESULTS

Staying ahead of the curve as gen AI use cases grow

As Cisco looks to expand gen AI even further, the possibilities are virtually endless. The company will continue to use MongoDB to support a wide range of intelligent applications, from improved customer support to greater automation and beyond. With MongoDB technology, Cisco can trust that its applications are secure, scalable, and high performing.

“AI is going to be integrated into pretty much everything,” says Santos. “It’s not a matter of what product will have AI, it is what product will not have AI in the future.”