A new startup has emerged aiming to augment network operations using AI.

Based in Richardson, Texas, NumoData is developing AI-powered software solutions capable of performing advanced troubleshooting to keep networks running smoothly.

Its platform can proactively perform actions like fault diagnosis and anomaly detection, while AI tools perform automated remediation, which the startup claims will help reduce network operational costs while ensuring both network performance and user experience.

At the helm of NumoData is Mary O'Neill, formerly Nokia’s VP of analytics. Having spent more than half a decade at the Finnish firm, she departed in May 2023.

O'Neill serves as the startup’s CEO, which formally launched this week, having been spun out of Exfo, the network equipment monitoring firm.

Private equity firm Teleo Capital is backing the newly independent company, as O'Neill and team look to bring its AI-driven technology to everything from radio access networks (RAN) to core and edge networks.

“With our commitment to developing AI and autonomous operations, we're building customer solutions that will define the next era of network intelligence,” O’Neill said. “Our mission is to empower operators – leveraging our trusted data to deliver smarter, faster, and more autonomous networks that enrich the digital lives of end users.”

“As a minority shareholder in NumoData, Exfo looks forward to supporting their continued success as they invest in the future of network operations and serve the evolving needs of communications service providers worldwide,” said Philippe Morin, CEO of Exfo.

NumoData joins an already stacked line of firms looking to apply AI to augment and support network operations.

AT&T, for example, has a generative AI-powered system dubbed Geo Modeler that uses synthetic data and a network foundation model to predict network coverage, dynamically changing coverage to where it’s needed most.

Carriers like Huawei and Ericsson have been exploring AI-powered autonomous network projects, while IBM launched an autonomous network solution back in September that uses foundation models capable of deciphering patterns from massive amounts of diverse time-stamped telemetry data.

And at the recent DCD Compute event in London, AI-powered processes like retrieval-augmented generation (RAG) were cited as ways for firms to instill network autonomy, pointing AI models at internal databases of network flows and telemetry to allow engineers to query network status in natural language.