A while back, I started tinkering with LLMs in my home lab, feeding them network configs to see what they’d spit out. The results were wild — one prompt, and I had a Python script for VLAN provisioning that almost worked. Another, and I got a decent root cause analysis for a flaky BGP session. But here’s the catch: as exciting as it was, it felt like playing with a wildfire. No guardrails, no audit trail, just raw AI power loose on my network.
That’s when it hit me: AI’s transforming automation faster than anything I’ve seen since the DevOps wave at Red Hat, but without security and enterprise-grade control, it’s a recipe for chaos.
That’s why I’m thrilled to announce the Itential MCP Server, launched today at AutoCon 3, where I’m sharing our vision live on stage. This breakthrough orchestration and governance layer, built on the Model Context Protocol (MCP), lets AI systems — from LLMs to AIOps agents — safely plug into Itential’s Platform, delivering secure, validated automation across your hybrid cloud infrastructure.
AI’s Promise Meets Reality
AI’s moving at a blistering pace — LLMs generating configs, AIOps spotting issues before they tank your SLA, agents reasoning through fixes faster than you can type “show run.” I’ve seen it firsthand, from my lab to customer war rooms, and it’s electrifying.
But here’s the problem: “smart” doesn’t mean “safe.” An LLM might suggest a firewall rule, but if it skips your change management process, you’re in trouble. An AIOps agent might fix a routing issue, but without an audit trail, good luck with your next compliance review.
Enterprises aren’t ready to let AI loose on production — not without serious guardrails.
And then there’s the data trap.
Years spent building a so-called “single source of truth” just to start automating, like constructing a fortress before fighting a battle. The industry’s preached this for years: clean your data, centralize it, then maybe automate. But AI flips this on its head. LLMs and agents pull data from anywhere — CLI outputs, cloud APIs, messy logs — and make sense of it in real-time, no centralized repository required.
This is where Itential shines.
Our platform, with the MCP Server at its core, is purpose-built to tame AI’s wildfire with unmatched integration prowess and enterprise-grade control. The Itential MCP Server, acts as the data plane for AI, normalizing data from CLI, APIs, or AI agents into structured, policy-enforced workflows. Whether you’re managing on-prem switches, AWS VPCs, or Azure firewalls, Itential orchestrates secure, scalable outcomes without the SSOT slog, letting you automate today, not after a data overhaul.
How Itential’s MCP Server Delivers
The MCP Server lets AI — whether it’s your favorite LLM, an AIOps agent, or a custom copilot — drive real infrastructure change, securely and at scale. It’s like giving AI a security badge and a playbook: it can act, but only within your rules.
Here’s how its transforming work for our customers:
- Human-Prompted LLM Workflow: You’re under pressure to fix a VLAN misconfig during a production outage. You prompt an LLM: “Generate a config to fix this.” The MCP Server pulls data straight from your Cisco and Juniper switches — normalizes it, and runs an Itential workflow through compliance checks and approvals. Within minutes, the config’s deployed, fully audited, and policy-compliant. One customer slashed their compliance review time by 60%, turning a multi-day slog into a quick win.
- Agent-Initiated Remediation: Your AIOps tool detects a congested link eating into your SLA. It pings an AI agent, which proposes a traffic reroute and calls the MCP Server. The server grabs telemetry and device data directly, orchestrates a multi-step workflow with Itential — validates, executes across cloud and on-prem routers, log it in ServiceNow — and resolves the issue before you finish your coffee. This cut MTTR by 70% for a telco handling thousands of daily alerts.
- Mid-Workflow Agent Invocation: You’re troubleshooting a mysterious BGP flap. The MCP Server kicks off a workflow, pulling logs from devices and cloud APIs, then calls an LLM to analyze them. The LLM pinpoints a misconfigured peer, suggests a fix, and the workflow validates it against your policies before applying it. This speeds up diagnostics by 50% for a financial services customer, with zero compliance violations.
These AI-driven workflows deliver measurable wins: accelerated compliance, slashed MTTR, and networks that run smoother under pressure. It’s automation that feels like a superpower, not a second job.
Why Itential’s Technology Was Built for This
I’ve been in the trenches — Cisco, Arista, Red Hat, now Itential — and I’ve seen automation evolve from clunky scripts to strategic platforms. AI’s the next leap, and Itential’s ahead of the pack. Our platform was architected to orchestrate across federated systems, normalizing data from CLI, API, and now AI/MCP into seamless workflows. Forget the old playbook of centralizing every scrap of data before automating — AI changes that, and our MCP Server makes it real. It lets AI agents pull and contextualize data from wherever it lives — device CLIs, cloud APIs, even messy logs — and our platform turns it into governed, scalable automation.
Security and governance are baked in, not bolted on. Every AI-driven action runs through policy checks, validations, and rollback safeguards, so you’re never explaining a rogue change to auditors. Our low-code UI lets practitioners build workflows fast, while CLI and API integrations give architects precision. Multi-agent collaboration ensures your AI tools work as a team, not a free-for-all. This is why some of the world’s largest networks trust Itential to deliver automation that’s enterprise-grade and ready for the AI era.
The Future’s Bright — & AI-Driven
The MCP Server is just the start. We’re infusing AI across our platform, from configuration management to lifecycle orchestration, to make automation smarter and more dynamic. Imagine natural language workflows where you tell your network, “Optimize this link,” and it happens — fully governed. Picture predictive insights catching config drift before it bites, or intelligent decision loops turning static scripts into living, reasoning systems. As AI evolves, clunky data repositories will become relics, replaced by agents that pull data dynamically — powered by Itential’s AI-driven orchestration. In 12-18 months, agent-to-agent orchestration will redefine how networks run, and we’re building the platform to lead that charge.
Join Me at AutoCon & Cisco Live
I’m at AutoCon 3 in Prague right now, sharing this vision on stage and hanging out all week to talk shop. If you’re here, come find me — let’s grab a coffee and brainstorm how AI can transform your network. Can’t make it? I’ll be at Cisco Live next, demoing the MCP Server and swapping ideas. Let’s build the future together, one secure workflow at a time.
A Shoutout to Our Partners
We’re not doing this alone. Our integration with Selector AI, showcased at AutoCon 3 and Cisco Live, is a perfect example: Selector’s AIOps platform detects anomalies, proposes fixes, and hands them to Itential for secure, closed-loop execution. Check out this blog to learn more.
Jump In Today
The Itential MCP Server is live here. Want to see it work? Watch our demo here to explore how AI and automation team up.
Want to just keep talking? Hit me up on LinkedIn or join the Network Automation Forum’s Slack. Let’s make AI work — securely, scalably, and without the 2 a.m. alerts.
Tags: AI CLI MCP Network Automation
Peter Sprygada
Chief Architect ‐ Itential
Peter Sprygada serves as the Chief Architect at Itential after serving as the Chief Technology Officer at Pureport where he was responsible for their multi-cloud network as a service interconnect platform. Prior to Pureport, Sprygada was a Distinguished Engineer for Red Hat, where he played the role of Chief Architect for the Ansible Automation Platform. Sprygada also held senior technical and leadership positions at Arista and Cisco, as well as several networking startups.
