Latest thoughts on Agentic Engineering
Agentic Engineering changes the economics of engineering bandwidth. Building is faster now: scripts, integrations, dashboards, tests, runbooks, and prototypes can move from idea to working draft in hours rather than weeks. That shifts the bottleneck away from writing code and toward knowing whether the output is correct.
In network engineering, that matters deeply. A generated config, automation workflow, or diagnostic model is only useful if it is safe against real topology, real vendor behaviour, real failure modes, and real operational constraints. The engineering craft moves toward sharper intent, stronger review, better test environments, and evidence-led validation.
The challenges are real. Company and customer information must stay protected, especially when engineers can accidentally paste sensitive network data, configs, or incident context into public models. We also need to help engineers feel safe as AI changes the shape of coding work. Their core value is not typing code; it is understanding the wider impact, risk, and interdependencies of systems and networks.
This is moving faster than anything we have seen before, and the field is nowhere near a steady state. I am keen to keep developing this space as a practical way to improve engineering efficiency, strengthen verification, and help teams build with more confidence.