You’ve likely heard the term agentic AI, and you may have a general sense that — at some point in the future — it will help automate your workflows. But Innovative Systems CEO Scott Alcott told Telecompetitor in a recent interview that it’s time broadband providers gained a deeper understanding of what agentic AI is and what it can do.
Alcott said he thinks most broadband providers understand assistive AI, which is useful for answering questions and combing through data, as well as generative AI’s ability to collect and process data to create code, charts, written pieces, and so on. He said most providers are currently using AI the way they might use an intern.
When it comes to agentic AI, Alcott said, people wonder “how is that different than any other plain old automation that we’ve had in broadband as an industry for decades?”
Alcott said he believes agentic AI adoption will grow strongly in the next 12 to 24 months, and that rural providers in particular are hungry for it. But he also thinks broadband providers need a more sophisticated view of agentic AI. To that end, he shared his three-point framework that clarifies why agentic AI is different.
1. How agentic AI processes data
Alcott said agentic AI harvests the data it requires when needed. “It is driven by events and conditions. It can query databases, it can pull logs and read documents, and it follows evidence rather than just fixed scripts and rules.”
Part of agentic AI’s power, he said, is how it handles unstructured data. “A field tech might write [in their notes] that the customer’s angry and they said they may want to leave. Yesterday, that was buried in logs, but tomorrow, that unstructured data will be read by agentic AI, and it will recognize it as a churn signal. It will carry that information — and all the ones like it — to a marketing team with a target list for saves and retention.”
Alcott also described agentic AI’s ability to use cross-domain reasoning — examining network performance, billing data, and device usage to let customers know they might be experiencing a problem or need a speed boost. For example, agentic AI could identify an app that is using lots of background data on a customer’s phone.
Agentic AI’s ability to harvest unstructured data from multiple platforms and use reason — even in the face of ambiguous data — is a game-changer for broadband providers. “We find use cases where it resolves device issues, billing issues, and customers being angry about their service. That depth, context, and root-cause analysis across a lot of signals makes agentic AI not just your grandma’s automation,” Alcott said.
2. How agentic AI operates
Agentic AI’s proactive nature is its second differentiator, Alcott said. Drawing on his previous example, he said, “Every six seconds, it can look at every customer and proactively figure out who is redlining their speed tier. If a customer has a device on their phone that is generating usage in the background, maybe that’s a security issue. Maybe it’s an app that is malicious. It can proactively notify them, not wait until they call and complain about their bill.”
Alcott described agentic AI as a “no-touch workflow” for broadband providers. “The customer who’s redlining their speed tier can be identified. The AI agent can quickly check [whether] the customer is eligible for an upgrade. Is there a higher speed tier? Does their device support a gigabit or two gigabits? Can their modem handle it? If so, it can push a notification: ‘You have an option for a promotion.’ To accept it, the customer can press a button, and it can activate provisioning and change their billing.”
Providers will have to decide the rules for when and how agentic AI notifies customers to avoid sending too many offers, annoying customers, or making them feel their privacy is being violated. But those necessary decisions don’t change the importance and power of this agentic AI feature.
“The ability to correlate vast amounts of data, look at unstructured information, work proactively — end-to-end — and close the deal is a very different thing than the automation that we used to have.”
3. What agentic AI produces
For broadband providers, agentic AI’s third differentiator is what it can produce. To illustrate, Alcott described how he recently used Anthropic’s Claude AI service to create an agentic employee named Stan — without any previous coding ability of his own.
“Stan scans the market, figures out when competitors make sales, if they’ve changed their offers, and if they’ve launched a new product. It does this review and automatically emails the product and marketing team,” Alcott said, describing Stan as “a marketing analyst in my house.”
The point, Alcott said, is that agentic AI can create lasting assets for broadband providers. “[Stan] lives in the company not just as knowledge or activities, but as a tool that is persistent and durable and getting smarter and more capable,” he said. “Agentic AI turns activities into durable assets that add value. It systemizes and builds permanent functions that are built into the fabric of the company.”
The path forward for providers and agentic AI: Two recommendations
How do broadband providers get started with agentic AI? Based on his experiences, Alcott offered two recommendations that he described as “controversial and important.”
First, Alcott said he thinks the move toward agentic AI should start from the top down. “Every CEO, especially if they have no background in technology, should be using these tools themselves, generating their own stance, and making it core to their companies,” he said. “It is adding value to employees, not subtracting. It’s creating new workflows and activities.”
And momentum shouldn’t only start at the top — Alcott also said that using agentic AI tools to enhance business productivity should become mandatory. He recommends estabilshing key performance indicators (KPIs) related to adoption and productivity.
Second, Alcott said that broadband providers must clean up their data for agentic AI to be successful. “There are studies that indicate that sometimes as [much] as 40% of the data has inconsistencies or ambiguities related to the real source of truth,” he said.
The problem is that when agentic AI processes inconsistent data, “it’s going to spin up snowballs of chaos” and create more problems than it solves. To make agentic AI successful, providers will need to invest in cleaning up their data and resolving any inconsistencies.
“The great news is agentic AI can help with that,” Alcott said. “It can wade through a decade of history and notice that there’s an inconsistency in the billing and field data. And based on the order of operations, it can tell you what is true.”
Alcott concluded by describing agentic AI as a positive development for broadband providers and their customers.
“It’s not always true that what is good for the operator’s efficiency, profitability, and lifetime value of a customer intersects with what’s good for the customer. But with the use cases we see with agentic AI, it’s a win-win-win-win-win. This is one of those beautiful moments where efficiency and benefits for the operator line up for a better customer experience and everybody wins.”
