Infovista unveils ‘pragmatic’ approach to telco AI

• Infovista has decades of telecom network software experience
• It integrated machine-learning capabilities into its products years ago
• Now it has developed an ‘agentic AI framework’ that the vendor’s CEO is pitching as a ‘pragmatic’ approach to helping telcos with their AI strategies

Infovista, one of the better known and respected names in the telco networking software sector, has joined the throng of tech firms with an AI pitch and plans to ‘show and tell’ at the upcoming MWC26 show in Barcelona, but its CEO is keen to stress that the Paris-based vendor is bringing a measured and “pragmatic” approach to the market, which is based on decades of network management software experience. 

Rick Hamilton, an industry veteran who has been Infovista’s CEO for more than two years and who has previously voiced his scepticism about how dramatically AI will impact telco operations, told TelecomTV that the company’s VistAI agentic AI framework has been developed to provide telcos with a range of tools that will be genuinely useful on a day-to-day basis. 

Infovista’s take on the current state of AI tools for telcos is that what is currently on offer from external suppliers presents “tradeoffs for operators. Generic platforms require extensive customisation, vendor-specific AI limits flexibility, and point solutions lack cross domain correlation.”

As a result, “we’re trying to be very pragmatic in what we’re releasing here, to [provide] what we think is going to be really important in the AI journey. We’re trying very hard not to start waving the flag and saying ‘We’re your AI salvation’, because there isn’t a vendor in the world that’s going to be anybody’s AI salvation. We all have meaningful roles to play,” noted the CEO. 

But some are more meaningful than others, according to Hamilton. “We know we can do better than anybody in the market, and this is what we’re going to focus on,” he stated. 

“Networking is complex – network data structures are complex. For 30 years we’ve produced some of the most compelling domain-specific network data that’s trustworthy, accurate, that’s focused on what carriers are trying to do. So we’re going to take advantage of that play because we know what data is out there in the context of the networks. Everybody has planning, network testing and network operations tools, but the depth and experience we have in all three isn’t easily replicated in the market,” said the CEO. 

That confidence is based on the telco network domain know-how that the Infovista team of about 600 staff has built from working with more than 1,000 telco and large enterprise customers that use the vendor’s network planning, optimisation, testing and assurance tools. Its customers include the likes of BT, Colt, KDDI, MTN, Orange, T-Mobile US, Telstra and Vodafone, and about 75% of its revenues come from its telco customers (mostly mobile operators). (The company doesn’t share its financial numbers but Hamilton said the company is “growing significantly faster than the markets we serve” and that its “operating leverage and profitability is what’s allowing us to invest in this agentic framework”.) 

So what is VistAI? It’s a key part of what Infovista is already pushing on its website as True Network Intelligence and is the result of 18 months of development that has seen the vendor’s R&D team use AI application development tools, such as LangChain and LangGraph, Chroma (open-source database) and large language models (LLMs), including OpenAI and Gemini, to develop the framework. 

It comprises about 90 discrete intent-based AI agents, one of which is an orchestrator, which enable operators to define outcomes – for example, ensure the quality of service (QoS) meets the service-level agreement (SLA) – that are then executed by the agents within configurable guardrails. The framework includes a natural language interface called VistAI Ask that enables network teams to query network data conversationally.   

The agents, all of which support model context protocol (MCP) for data queries and agent-to-agent (A2A) for AI agent communication and collaboration management, enable a range of automated network management tasks, including multi-domain correlation across core, transport and access architectures, and have been developed mostly based on customer feedback. 

Hamilton says the agents are available for Infovista customers to use today and are already embedded across the vendor’s four product sets. “We have real-life experience with two big operators around the world, and we have just gone through a validation with one very, very big enterprise in the US… we’re trying to validate at the coal face… [and] we’re bullish about what we’re seeing with active customers today, and the feedback they’re giving us.”  

And the message that Infovista is giving its customers is that they “have a licence, they have a relationship with us, they have active maintenance and support, now it’s just a matter of moving forward… [we don’t need to] do a hard sell. We’re talking to customers about how to take advantage of the technology that is already [embedded] in the products they get from us. And our hope is that we can demonstrate that capability across multiple domains, and that’s what will grow our business. We’re not trying to monetise the hype of AI. We want to make it real and pragmatic and effective, and that’s what has kept Infovista around for 30 years.”

Hamilton added: “AI is a very exciting thing, but it’s not changing our strategy as a company. We’re still doing what we’ve been doing forever. We’re just doing it in the context of evolving  technology. We have a duty and an obligation to play in this in a way that helps our customers get the most out of it… without throwing a bunch of money down the drain, chasing an AI thing that’s fairly immature at the moment.”

- Ray Le Maistre, Editorial Director, TelecomTV