One of the great sporting cliches is that the best teams do the basics well. Only once those basics have been mastered can you concentrate on the outcome. Doing the basics well doesn’t mean the same thing as doing the simple things well. Basics aren’t simple – mastering the basics requires discipline, patience and a deep understanding of the fundamentals of your role.
Right now, the industry conversation is full of promises on the integration and adoption of AI technology in network operations: AI-native networks, intelligent OSS, agentic automation, and autonomous operations. Pressed by strategic priorities such as investor pressure, the “race” for AI relevance, the techco transformation, there’s an understandable focus on the outcome of AI-native networks.
Joe Cumello, is SVP and GM of Blue Planet, the telecoms software business within Ciena. Last year Blue Planet revenues grew 49% year on year to $115M, driven by sales of inventory, orchestration and assurance software that are supporting the move towards automated networks. As our piece, AI and the fight for data control reported last year, the company also introduced its agentic AI Studio in 2025, and already has wins in that area.
Cumello is sure that AI-driven autonomous networks represents a shift in industry paradigm to rank alongside previous shifts such as SDN, virtualisation and cloud. But he cautions that delivering on that promise will require a focus on the basics.
“Success won’t come from marketing slogans or ambitious announcements. It will come from fundamentals: trusted data, secure architectures, and platforms that give operators control.”
The first of those fundamentals is generating and accessing clean, context-aware data. Cumello says that before operators can deploy advanced AI agents or autonomous workflows, they need to confront a long-standing problem – data quality.
“AI is only as good as the data you’re working from,” Cumello said. “You can announce all the AI platforms you want, but without clean, context-sensitive data in the domains you operate in, none of it works.”
As Cumello tells it, over the years, many telecom companies have centralised their information into modern data lakes. That’s a useful first step. But simply aggregating data isn’t enough. Telecom environments are highly interconnected, and AI systems need to understand those relationships.
That’s where context-aware data comes in. This matters because legacy systems often contain inaccuracies. Old inventory platforms may still list closed sites, retired equipment, or “ghost” infrastructure. If AI acts on bad information, it can produce costly mistakes, such as dispatching truck rolls to the wrong locations or misallocating resources.
But it’s not just about knowing what assets exist. Context-aware data means understanding how services connect across optical, IP, and mobile layers, how SLAs link to infrastructure, and how systems interact operationally.
Secure the Foundation
Security is another foundational piece – a basic – that can’t be an afterthought, especially in autonomous environments.
As networks become more self-operating, the risk profile changes. AI agents gain the ability to make operational decisions automatically. That creates new opportunities not just for efficiency, but for potential misuse.
“If a threat actor can insert a bad agent or corrupt the data feeding those agents, you could end up with network outages or ransom-style scenarios,” Cumello says.
The goal isn’t to slow innovation, but to design security into the architecture from the beginning – ensuring clear controls over who can deploy agents, how they’re validated, and what guardrails govern their actions.
There is often more talk than action around security-by-design approaches and Cumello admits the industry hasn’t solved this completely yet, but the right move is to address it upfront rather than bolt it on later.
Democratise AI, Don’t Recreate Old Models
A third necessary shift that Cumello describes is what he termed the democratisation of AI.
Historically, OSS/BSS environments have relied heavily on professional services and custom development. Vendors built complex systems that only they – or expensive integrators – could modify. Change requests became a revenue stream.
But operators increasingly want out of that model.
“They don’t want to be beholden to suppliers for every change,” he says. “They want to democratise decision-making and development.”
That means giving internal teams the tools to build and modify automation themselves. According to Cumello, Blue Planet’s own approach has always been to productise OSS capabilities and layer AI tools – such as low-code or no-code studios – on top. Customers can access clean data, select or build agents, and design automations without constant vendor intervention.
“If customers want us to do the work, we’ll do it,” he adds. “But the point is to empower them, not lock them in.” In this model, vendors provide the platform and guardrails; operators and integrators build the intelligence.
Financial Pressure
While the technology is compelling, the real motivation for AI transformation isn’t innovation for its own sake. It’s financial. Telecom operators face slowing service revenue growth and increasing operational costs. Investors and boards are demanding better margins. Autonomous networks offer a path to both faster service delivery and significant cost reduction.
“The conversation isn’t about deploying AI because it’s trendy. It’s about taking hundreds of millions, or billions, of dollars out of operations while improving agility.”
This connection between AI and measurable financial outcomes has fundamentally changed customer discussions, Cumello said. A year ago, many projects focused on replacing aging inventory or assurance systems. Today, operators are asking bigger questions around redesigning operations to reduce cost at scale, automating end-to-end processes, and eliminating legacy dependencies entirely.
How the Industry Is Moving Forward
Despite the seeming never-ending drumbeat around AI, adoption itself within the telco operations space is pragmatic, Cumello said. Operators aren’t attempting massive, overnight transformations. Instead, they’re working use case by use case to clean the data, automate a manual process with an agent, measure the savings and expand from there. The approach is deliberate rather than experimental – focused on real operational improvements rather than proofs of concept.
“It’s incremental,” Cumello says. “You check that the data is clean, confirm you’re getting the benefit you expected, and then grow.”
He adds that Blue Planet already has paying customers deploying these capabilities, with more in progress. The approach is deliberate rather than experimental – focused on real operational improvements rather than proofs of concept.
Expect to hear more on this at MWC Barcelona and after.
