Itential’s CTO outlines why telco AI hinges on expertise
57th article in the last 90 days, one of 224 articles referencing Itential. Previous coverage: Edge AI networking, agent platforms, and security advisories - Week of April 27, 2026 (Apr 2026).
Companies mentioned
Best suited for
- Seniority
- Director
- Job function
- Non-Technology / Industry Analyst
- Buyer role
- Decision Maker / Budget Holder
- Decision stage
- How to start
- Adoption curve
- Early Majority
- Technology maturity
- Market Correction
- Industry
- Telecommunication Services / Diversified Telecommunication Services / Integrated Telecommunication Services
Our classification, not the publisher's statement. Best suited for, not only for.
Itential CTO Chris Wade argues that telco’s AI challenge centers on expertise transfer and infrastructure architecture rather than on AI model quality, stressing programmability and open, documented APIs so agents can act on existing network operations under governance controls.
Research Overview
Wade frames telco network automation as a matured capability with REST APIs, NETCONF, YANG models, and orchestration platforms that have improved over time. He says the historical barrier has been where expertise sits and how it is converted into automated workflows.
He also describes how AI changes the cost of encoding procedures by allowing agents to reason through method documents, design specifications, and runbooks. The result, in his view, is broader organizational access to operational knowledge that previously depended on a small set of engineers.
Key Findings
Wade says telco organizations have long faced an “expertise problem” because automation work required developers, documentation cycles, and months of effort. He adds that teams created to handle this conversion became bottlenecks, keeping knowledge locked in limited locations.
He argues that AI reduces the transfer cost so expertise can be packaged into workflows that the wider organization can use. He characterizes this as a structural shift in how telco operations scale when that knowledge becomes accessible to agents.
Technical Breakdown
Wade contrasts two common AI frames for network operations: chatbots that answer questions and fully autonomous agents that make changes without human involvement. He says chatbots are too limited and that fully autonomous agents are too risky for telco networks.
Instead, he recommends a layered architecture that uses deterministic execution for cases where determinism fits and adds reasoning for edge conditions. He also emphasizes building a governance layer before expanding agent permissions and keeping the “happy path” mostly deterministic.
Operational Impact
Wade argues that telco vendor ecosystems have historically treated API access as proprietary, using closed interfaces, proprietary models, and integrations that depend on vendor professional services. In his description, this model conflicts with an agentic approach because an agent cannot operate on systems it cannot reach.
He says operators moving fastest require API accessibility as part of procurement and assess whether their AI can reach vendor systems directly. He also outlines a transition approach focused on observation and recommendation first, paired with audit trails that show what an agent did and why.
He adds that teams that move directly to autonomous action can revert after incidents, resetting the program timeline. He concludes that the work over the next three years is building the automation layer, data layer, and governance layer that lets AI act on what telcos already have.
Blog post, originally published by Chris Wade at itential.com.