By Douglas Wadkins
Major advances in technology come in waves, and the next one, driven by Artificial General Intelligence (AGI) underpinned by quantum computing, may be the most disruptive we have ever seen.
Within the next five years, trustworthy digital assistants will become common. Once that has happened, the next phase will be autonomous, agentic, and eventually AGI driven. Instead of simply replying to prompts, agentic systems will take goals, call tools, traverse APIs, and collaborate with other agents.
These agents will work 24/7, never give up, and act in fractions of a second, so architecture and skills will have to adjust. AI assistants will absorb more routine coding, but only if architectures keep concerns separated and codebases manageable. Agentic systems in operations will need dependable connectivity to services across regions and clouds, while product and architecture roles will continue to converge into hybrid profiles that understand system behaviour and network constraints together.
For boards and regulators, this will increasingly look like a combined AI and quantum risk, pulling network design, management planes, and governance systems into the centre of the discussion. One option is to build an agentic governance channel on top of the management network.
Data from management network edge devices would travel over a separate, secure network into a shared data store, where it can be used for retrieval to ground agent decisions, with any model training data handled under separate controls. Access for AI agents will need to be scoped, time-bound, and explainable.
Any change the agent wants to make would have to pass a policy check on the management layer before it is approved, carried out, and, if needed, reversed. Every proposal, action, or access attempt will need to be logged for governance and compliance purposes.
With the rate of change in this area, boards need to prioritise policy development in 2026.
The author
Douglas Wadkins is Chief Technology Officer, Opengear
