Mist Systems Inc., a high-profile networking startup backed by Alphabet Inc.’s GV, wants to automate enterprise Wi-Fi infrastructure using artificial intelligence.

Mist sells a platform called Learning WLAN that enables companies to build what it describes as “self-learning” wireless networks. The offering comprises an AI-infused management service and a family of Wi-Fi access points. Today, the startup introduced more AI features for the management backend to ease administrators’ work.

The first change is an enhancement to Marvis, a virtual assistant that powers some of Learning WLAN’s diagnostics features. Mist said the bot’s natural-language processing interface has been expanded to recognize hundreds of different request types. Administrators can pose questions such as “How is the access point in the conference room doing?” through a section of the management service’s console that resembles a search bar.

Besides fetching the information, Learning WLAN can uses it to make configuration changes. That’s thanks to a new automation feature Mist is introducing alongside the Marvis update. The capability is designed to help with radio resource management, which is the task of fine-tuning parameters such as wireless signal strength to optimize coverage.

According to Mist, Learning WLAN tracks key network metrics and automatically tweaks settings when requirements change. Administrators can specify what operational conditions the system should strive to maintain at any given time by defining “service level expectations” through the management console.

This latter mechanism has other uses as well. Thanks to the new update, administrators can now also set service level expectations for the wide-area network that connects Wi-Fi access points with the rest of the company’s infrastructure. According to Mist, the enhancement makes it easier easier to determine if a wireless connectivity issue stems from a problem with a different part of the network.

Rounding out the update is an upgraded anomaly detection mechanism. Mist said that its platform now uses deep learning to detect unusual events and items which may potentially warrant special attention.

Image: Pixabay