From AIOps to Autonomous Networking
AI isn't a feature, it's an organizational strategy. Juniper Networks is not only embedding AI into the platform (Marvis), but also into their support structure and company culture. Ticket deflection, root cause analysis, and issue resolution are becoming part of a closed-loop learning system.
Other organizations need to take note. AI is not as simple as putting it into your strategy. You must fully integrate it into your workflows and company culture to back up your AI story.
Marvis Actions are now moving from suggestive to corrective. For example, Marvis can detect a missing VLAN and recommend a fix. But now, it will be able to apply the fix for you. What drove this shift was enforcing domain expert reinforcement learning between the Mist Data Science Team, Mist Customer Success, and the Customer. AI lives through these teams.
There's a direction toward self-driving actions with permissioned change execution. That is the next step for AI to take action on the network. Marvis has been learning what is wrong with the network. Marvis then began making suggestions when issues occur. Now, Marvis will begin taking action to resolve issues.
Juniper Networks has always had it in the plan to work towards a self-driving service for the network. Bob Friday has shown this in previous Mobility Field Days. And today we are one step closer to seeing it become reality.
Juniper Networks introduces the GenAI Conversational Interface which isn’t just a chatbot. It mirrors tools like ChatGPT, letting engineers query the network directly to build their own dashboards. This is moving towards Agentic AI. The Large Language Models help feed the autonomy of the Marvis conversational interface, or shall we call it MarvisGPT?
These developments signal that operational roles in networking will become increasingly AI-augmented. Get ready for network engineers to pick up AI skills. Today, learn how to write prompts that generate the results we need. But eventually, we will have roles bridging networking and AI.
