When I set out to evaluate how AI could improve IT operations, I didn’t write a whitepaper. I built a complete enterprise environment to test the thesis firsthand , and learned as much about where AI falls short as where it delivers.
The Question Every IT Leader Is Facing
Every IT leader I talk to is facing the same question in 2026: “What should we actually do with AI?”
Not the vendor pitch version. Not the consultant slide deck. The real, operational answer , the one you’d stake your infrastructure on.
After 30 years in IT leadership , from Marine Corps avionics to enterprise hospitality to maritime fleet operations. I’ve learned that the only way to answer a question like that is to build the answer. So I did.
What I Built
Over the past several months, I’ve constructed a complete enterprise-grade infrastructure environment from scratch. Not a toy lab. Not a sandbox. A fully operational environment with the same architecture patterns, monitoring, automation, and security controls you’d find in a production enterprise.
The hardware: A 5-node Proxmox cluster (40 cores, 320GB RAM, 10TB NVMe) running 15 virtual machines. Three NICs per node with dedicated management, VM traffic, and storage networks.
The software stack: NetBox for IPAM/DCIM as the source of truth. Ansible for configuration management across 16 hosts. Naemon for monitoring with 65 monitored hosts. Loki and Grafana for centralized logging across 13 hosts with 8 dashboards. pfSense for firewall and network segmentation. Vaultwarden for credential management. MediaWiki for documentation with 35+ pages and formal SOPs.
The integration layer: 8 custom API servers built on the Model Context Protocol, providing 336+ tools that give the operations platform direct, authenticated access to every system in the environment. 20+ custom skills for common operations.
The Thesis: API-First Operations
Most organizations are experimenting with AI as a chatbot, a code assistant, or a document summarizer. Those are useful, but they’re not transformative.
The transformative use case for IT operations is an integrated platform that can query your IPAM, check your monitoring, read your runbooks, execute commands over SSH, manage your firewall rules, and update your documentation. Not in theory. In practice, with real authentication, real error handling, and real audit trails.
That’s what I built. And then I measured what happened.
The Results: 7-10x Productivity Multiplier
I track every working session with structured efficiency logging , time spent, task category, estimated manual equivalent, and calculated productivity ratio. After 48 sessions across multiple months, the pattern is clear:
- Average productivity multiplier: 7-10x compared to manual execution of the same tasks
- Peak sessions exceed 15x for tasks involving multi-system coordination (deploy a service, configure monitoring, update documentation, add to inventory , all in one pass)
- Infrastructure built in months, not years , the complete environment from bare metal to fully monitored, documented, and automated
These aren’t estimates. They’re measured. Every session has a log entry with actual duration versus estimated manual duration.
The ROI Model That CFOs Love
Here’s the number that gets the most attention: the entire software stack costs $240 per year. That’s it. Domain registration and a couple of small subscriptions.
The commercial equivalent , Datadog for monitoring, ServiceNow for ITSM, SolarWinds for IPAM, HashiCorp Vault for secrets, Confluence for documentation , runs $75,000 to $109,000 annually at comparable scale.
This isn’t an argument against commercial software. It’s a demonstration that an IT leader who understands the full stack can make strategic build-vs-buy decisions that directly impact the bottom line. That’s a CIO skill, not just a technical one.
What I Learned About AI in Operations
After building and operating this environment, here’s what I’ve learned about using AI as part of an operations platform:
1. The infrastructure has to come first. AI is only as capable as the tooling underneath it. Without proper APIs, a source of truth, and structured documentation, AI is just a fancy text generator. The API servers, the NetBox-as-source-of-truth pattern, the SOP framework , those are what make AI actually useful in operations.
2. The multiplier comes from multi-system coordination. The biggest productivity gains aren’t from any single task running faster. They come from a platform that can touch six systems in one operation: create the VM, add it to inventory, configure monitoring, update the wiki, set up logging, and commit the changes. That coordination is where human time gets consumed and where automation excels.
3. Operational discipline matters more, not less. Working with AI-augmented tools requires better naming conventions, better documentation standards, and better operational procedures than working alone. The SOPs, the credential management standard, the documentation architecture , these aren’t overhead. They’re what make automation reliable.
4. This is a leadership skill, not just a technical one. The ability to design the tooling strategy, structure the automation patterns, and measure the outcomes , that’s strategic IT leadership. The executives who understand how to evaluate, govern, and deploy these capabilities will have a significant advantage as AI-augmented operations becomes mainstream.
What’s Next
I’m continuing to build on this foundation. Current projects include completing a pfSense migration for full network segmentation, deploying 802.1X network access control with automated device onboarding, and expanding the case study documentation to share specific implementation patterns.
If you’re an IT leader evaluating AI for operations, my advice is simple: stop evaluating slide decks and start building. The gap between “AI-curious” and “AI-operational” is where the next wave of IT leadership differentiation will happen. And it’s a gap you can only close by doing the work , and being honest about where AI helps and where it doesn’t.
Craig Corbin is an IT executive with 30+ years of progressive leadership spanning military operations, enterprise hospitality, and maritime fleet technology. He specializes in enterprise infrastructure architecture and AI-augmented operations. Connect on LinkedIn or visit the Technology Lab to learn more.