There was a time when a small ISP could rely on a handful of people who understood the entire stack. Network. Systems. Voice. Security. One person could troubleshoot across all of it. Not perfectly, but effectively.
That model has become much harder to sustain.
Over the last decade the industry moved toward specialization. Training programs, certifications, and career paths all trended in that direction. It made sense. The enterprise market drove most of the demand and enterprise environments needed deep expertise in narrow domains. A network engineer focused on routing and switching. A systems administrator focused on servers and virtualization. A security specialist focused on threat detection and response. All highly skilled. All necessary.
But that shift created a problem for smaller operators that nobody planned for.
The small operator reality
Earlier in my career I held a role that required broad capability across voice, network, systems, security, and virtualization. Not surface level familiarity. Working knowledge across all of it. That combination was hard to develop then and it is significantly harder to find now.
The reason is straightforward. The industry largely stopped producing people with that profile.
For a large enterprise that is not a crisis. They can hire four specialists and build a team with full coverage. For a small operator running a technology team of two or three people, that is not an option. The economics do not support it. You need one person who can cover the ground, and the pool of candidates who can do that has been shrinking for years.
This is not a criticism of specialization. Depth has real value. When something breaks at 2am in a complex environment you want someone who has seen every variation of that failure. But breadth has value too. Pattern recognition across disciplines. The ability to see how a change in one layer affects three others. The generalist is not a compromise in a small operator environment. The generalist is a strategic necessity.
The problem is that "jack of all trades, master of none" became the dominant narrative, and the industry responded accordingly.
Where AI enters the conversation
Over the last few months, I have been wondering if AI could help solve this problem.
Not as a replacement for technical expertise, but as a force multiplier for the generalist. Consider a network administrator who understands the fundamentals of cybersecurity but has never configured a SIEM or worked through an incident response process. Historically that gap meant calling a specialist or a vendor. Today that same administrator can work through an unfamiliar security event with AI assistance, ask the right questions, understand the output, and execute a response, because the foundational knowledge is there and AI can bridge the rest.
The generalist provides the context. AI provides the depth on demand.
Some would argue that AI needs to know the specifics of your network to be truly useful. In some cases that is true. Detailed topology, device configurations, and vendor specific quirks matter when you are troubleshooting a complex outage. But a lot of the value does not require that context at all. Knowing how to approach an unfamiliar security event, interpreting log output, working through a configuration problem on an unfamiliar platform, none of that requires AI to have a map of your network. The foundational knowledge transfers regardless of environment.
The more interesting question
But here is where it gets more interesting. What if AI became more than a problem-solving tool? What if it became the institutional memory of the network itself?
Small operators have always struggled with knowledge transfer. A senior technologist retires or moves on and takes 20 years of undocumented tribal knowledge with them. The next person starts over. That cycle is expensive and it never fully gets solved because documentation is always the last priority when the team is already stretched thin.
AI changes that equation. Configuration decisions, troubleshooting histories, network specific quirks, vendor relationships, all of it can be captured, organized, and made accessible in a way that static documentation never achieved. Not just stored, but searchable. The next technologist does not inherit a folder of outdated Visio diagrams. They inherit a working knowledge base they can actually have a conversation with.
If that becomes the model, AI is not just a force multiplier for the generalist. It becomes the institutional foundation that small operators have never been able to build. The more a technologist understands their own environment, the more effective the AI assistance becomes. And you do not need perfect documentation to start. You build the knowledge base as you go, and it compounds over time.
The question worth asking
The generalist vs. specialist debate has been framed as a tradeoff for as long as I can remember. Depth vs. breadth. Expertise vs. flexibility. Jack of all trades vs. master of one.
AI may be the first development that actually changes the terms of that tradeoff rather than just forcing you to pick a side.
Is anyone seeing this play out in practice? I would be curious to hear from other small operators on whether AI is starting to fill this gap, or whether we are still in the theoretical stage.