Analyst Report

Gartner Predicts 2026: AI Agents Will Reshape Infrastructure & Operations

Agentic Operations Are Coming Fast. Governance Decides Who’s Ready.

AI agents are moving past chat and copilots into real infrastructure operations. Gartner predicts AI will evolve “from tools that assist humans to platforms that replace manual effort for complex workflows.” The hard part is not getting an agent to recommend a change. It is trusting it to make one on production infrastructure.

Get the report to see what Gartner predicts is next, and what separates teams that scale agentic operations from teams that stall.

The AI Operating Model

The Operating Model for Infrastructure Is Changing

I&O teams are under pressure to move faster with fewer people, while infrastructure gets more distributed, more hybrid, and less forgiving. Gartner predicts AI agents will take on more of the planning, execution, and continuous optimization that used to sit with engineers, and that the I&O role will shift from doing the work to supervising the agents that do it.

The interface changes too. Gartner expects I&O’s interaction model to move from CLI and scripts to “prompt engineering, policy definition, and workflow orchestration,” as teams frustrated by the limits of traditional automation turn to agentic AI they can direct in natural language.

This is a change in operating model, not just tooling. And a system only earns the right to act when every action it takes is governed.

Autonomy is the easy part. Trusting it in production is the hard part.

Key Findings From the Report

- AI is evolving from assistants to autonomous agents that plan, decide, and fulfill goals, moving the market from tools that assist humans to platforms that replace manual effort.
- Human involvement in agentic workflows drops sharply as autonomy rises.
- Agentic NetOps is more scalable and cost-effective than manual operations and outsourcing, pushing enterprises from managed services toward DIY.
- As agents take on mission-critical roles, performance SLAs, audit, and accountability become mandatory.
- I&O leaders will adopt agentic AI to automate complex, nondeterministic operations through a natural-language interface.

Gartner’s Take

Gartner’s Take: Autonomy Without Control Is a Liability

Gartner’s Strategic Planning Assumptions for agentic AI in I&O:

70%

of enterprises will deploy agentic AI as part of IT infrastructure operations by 2029, up from less than 5% in 2025.

40%

Human-in-the-loop falls to 40% in IT operations workflows by 2028, down from 95% in 2025.

50%

of I&O organizations will be reshaped around AI agents by 2030.

25%

overspend on network management by 2030 awaits organizations that fail to adopt agentic NetOps.

The through-line is simple: autonomy goes up, humans step back, and budgets follow. None of it works without a way to turn an agent’s decision into a safe, governed action, held to the same approvals, RBAC, and audit trail you would demand of any human change. That is what agentic operations means at Itential.

AI adds reasoning. Itential adds the guardrails.

Itential provided the orchestration backbone – governance, velocity, and scale – to make safe autonomy practical at enterprise level.

Greg Freeman

Vice President Network and Customer Transformation, Lumen

The Difference

Why Most AI Agents Won’t Survive Production

Generic AI platforms are good at automating business apps. Infrastructure is less forgiving. A wrong change doesn’t return a bad answer. It takes down a network, breaks compliance, or starts a Sev1 at 2am. The difference between an agent that demos well and an agent you can actually run comes down to how it acts.

An Agent Demo	An Agent in Production on Itential
Touches systems directly	Acts only through the governed platform, never the systems directly
Permissions assumed	Scoped tools locked at design time, validated at runtime
Probabilistic execution	Deterministic execution across three modes, one engine
No record of what changed	Immutable audit for every action, human or AI
One vendor, one domain	Multi-vendor and multi-domain by design

Gartner tells I&O leaders to scrutinize exactly this: to demand genuine planning and execution autonomy instead of rebranded assistants or RPA, and proof tied to production rather than demos. On Itential, FlowAgents act through governance built in, not bolted on and deterministic, real execution, so AI moves at machine speed inside the controls you set.

AI recommendations are easy. Safe AI execution in production is not.

How Itential Delivers

What Gartner Says to Demand. What Itential Delivers.

Itential is the agentic operations platform for infrastructure. Gartner’s guidance for adopting agentic AI safely reads like a checklist. Here is how Itential answers it.

What Gartner tells I&O leaders to demand	How Itential delivers
An AI agent engineering platform to manage agent permissions, integrations, testing, and guardrails	Scoped tools locked at design time, two-layer agent RBAC, and 1,000+ integrations, on one platform
Governance mapped to change controls, with approval ladders and rollback	Governed change management with pre-checks, post-checks, approval gates, and rollback on every action
Observability and audit logging for every agent action	Immutable audit for every action, human or AI, with full attribution
Genuine planning and execution autonomy, not rebranded assistants	FlowAgents built with FlowAI reason through goals and take governed action, they don’t just recommend
Deterministic, production-grade execution, not demos	Deterministic execution across three modes, one engine, proven at carrier scale

Two things Gartner doesn’t have to tell you, because they are how Itential was built. It is purpose-built for infrastructure, multi-vendor and multi-domain across hybrid and multi-cloud environments. And it keeps zero copies of your infrastructure data, so your systems of record stay the source of truth.

Same governance. Every action. Human or AI.

Itential Customers

What Agentic Operations Looks Like in Production

Gartner tells buyers to demand production proof and talk to reference customers. Here is what agentic operations looks like when it runs in the most demanding infrastructure environments in the world.

Lumen collapsed 1 billion alerts into 57,000 actionable incidents and runs 350+ live workflows, on a path to 80% machine-to-machine operations. Governed autonomy at carrier scale.
Lumen cut customer-impacting incidents by 21%. Autonomy that reduces risk instead of adding it.
Southern California Edison removed 90% of the manual steps from network change, without giving up control.
A global Tier 1 telco put 5 FlowAgents into production in 4 days. First value in days, not months.

Generic tools execute. Itential remembers what changed.

AI Ready

How to Get Ready for Agentic Operations

Gartner’s advice is to start now, start safe, and govern from day one. Three moves put that into practice.

1. Govern first, then grant autonomy.

Stand up approval gates, RBAC, and rollback so every action is controlled before you scale. Start with governed change management.

2. Start with low-risk, high-volume work.

Gartner recommends beginning with well-defined, high-volume tasks like alert deduplication, triage, and known-error remediation. Build your first FlowAgent there, where execution is scoped, validated, and audited by default.

3. Consolidate before you scale.

Gartner expects agent platforms to absorb tool sprawl. Bring scattered scripts and tools into one model, deliver infrastructure as a product, and make it part of infrastructure modernization.

From I&O leaders to platform and automation teams, Itential is the orchestration foundation for adopting agentic operations faster, safer, and with control.

Gartner, Predicts 2026: AI Agents Will Transform IT Infrastructure and Operations, Cameron Haight, Ashish Banerjee, Joe Antelmi, Jonathan Forest, Chandra Mukhyala, 4 December 2025.

Gartner does not endorse any vendor, product or service depicted in its research publications and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s Research & Advisory organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.

Ready to Put AI Agents to Work on Real Infrastructure?

See how the Itential Platform turns agent reasoning into governed action across your network, cloud, and IT systems.

Request a Demo

The Latest in Agentic Operations & Infrastructure

Blogs

Not Every Agent Is Agentic: The Difference That Actually Matters

Gartner® Market Guide for AI Assistants for Infrastructure as Code

Guides & Whitepapers

AI Agents for Infrastructure: A Practical Taxonomy of Agent Types, Paradigms, & When to Use Each

Guides & Whitepapers

Agentic Operations for Infrastructure

Guides & Whitepapers

From Automation to Agentic Operations: The Practical AI Adoption Guide for Infrastructure Teams

Frequently Asked Questions

AI Agents in Infrastructure & Operations

–+

Agentic AI for infrastructure operations is AI that can plan, decide, and take action on IT systems, not just answer questions or make recommendations. Unlike assistants and copilots, an agent pursues a goal and executes the steps to reach it. On Itential, that means agentic operations: AI reasoning paired with governed execution on real infrastructure.

–+

Agentic NetOps is the use of goal-driven AI agents to run network tasks and processes with little or no human in the loop, moving people from doing the work to supervising it. Gartner expects it to be more scalable and cost-effective than manual operations or outsourced managed network services.

–+

An assistant responds. An agent acts. Assistants and copilots answer questions and suggest next steps, but a human still does the work. An AI agent plans toward a goal, chooses the tools, executes, and reports back. That shift from assistants to autonomous agents is exactly what Gartner describes in this report.

–+

Gartner predicts a fast move from AI that assists humans to AI that replaces manual effort for complex work. Its headline assumption: by 2029, 70% of enterprises will deploy agentic AI as part of IT infrastructure operations, up from less than 5% in 2025. The full report covers the role, budget, governance, and vendor changes that come with it.

–+

The same way they govern any change: role-based access, approval gates, audit logging, and rollback, applied to every action regardless of who or what triggered it. Itential enforces this through governed change management, so an agent can only ever do what policy allows.

–+

Yes, as long as they never touch the systems directly. On Itential, agents don’t get raw access. They act through the platform, where execution is scoped, deterministic, and audited. Reasoning can be probabilistic; execution cannot. That separation is what makes autonomy safe enough to run in production.

–+

It is the platform layer that connects AI agent reasoning to safe, governed action across infrastructure, managing permissions, integrations, guardrails, and execution in one place. Gartner calls this an AI agent engineering platform. Itential is the agentic operations platform for infrastructure, built for exactly this.

–+

FlowAI is Itential’s reasoning layer for agentic operations. FlowAgents built with FlowAI reason through a goal using live infrastructure context and a scoped set of tools, then act through the path that fits the job: automations and scripts via Itential Gateway, direct API calls, or platform workflows. Deterministic, agentic, and hybrid execution on one engine.

–+

Not replace, reshape. Gartner expects the I&O role to shift from doing tasks to supervising the agents that do them, with the highest value moving to people who teach, govern, and handle the hard exceptions. Deep domain expertise becomes more important, not less, because it is what trains and constrains the agents.

–+

Start small and governed. Pick a low-risk, high-volume task like alert triage or known-error remediation, put approval gates and rollback in place, and build your first FlowAgent there. Prove it in production, then expand. Consolidating scattered scripts into one model as part of infrastructure modernization keeps agent sprawl from taking hold.

Home
Resources
Analyst Reports