Enterprises Don't Have an AI Problem. IT Has an Absorption Problem.

In all the chatter about AI technology, there’s very little mention of the core issue: that enterprises aren’t getting the value they expected because they’re stopping at the adoption stage.

If you’re simply glancing at headlines or overseeing implementation at your own company, it may seem that enterprise AI adoption is going along smoothly. Plenty of research has been done on AI adoption in IT and growing budgets for AI technology, such as a recent Goldman Sachs survey of CIOs, which found that 42% of respondents expect AI to exceed 10% of their budgets in the next three years. 

While businesses are allocating budget to AI technology and pilots, there’s still a gap between news of initial adoption and news of innovative outcomes. It’s less clear why AI pilots fail, or why realizing AI ROI often presents a challenge. This is the gap between acquisition and absorption, where new AI technology comes on board, but the IT operating model barely changes. 

The constraint isn’t the model or vendor choice. Many organizational failures to find value with AI are due to the business’s capacity to absorb autonomy. Finding value in AI requires enterprise absorption, going beyond the first steps of acquisition and adoption.

The procurement illusion: why enterprise AI adoption feels like progress

It’s become quite easy to check off the AI box by procuring new AI-driven technology for the organization. Every IT service management (ITSM) vendor is shipping something they call AI. But acquisition doesn’t equal adoption, or absorption. Too many IT projects stall because enterprises stop at adoption rather than pushing through to absorption. 

There are generally three stages in this pipeline:

Where AI actually stalls inside IT teams

For autonomous IT transformation specifically, the absorption roadblock takes on a few different facets. First, too often, workflows remain the same, even after adopting AI tools, which means that every help desk ticket still moves through human queues, built for a system of record vs. a system of action.

Next, the trust threshold of AI can create a big roadblock. Without defined guardrails for AI usage, particularly for agentic AI implementation, the AI tools that were adopted simply stay in an advisory role, without the ability to act. 

And metrics and incentives for adopted AI often don’t evolve to align with the new tools. IT technicians still get evaluated on tickets closed, rather than tickets eliminated. 

Finally, adopting AI without getting to the absorption stage ends up as tool sprawl. When teams bolt AI onto legacy ITSM, there’s no context for it to act. Without embedding AI technology, it can only serve as a suggestion engine.

The real bottleneck is the operating model, not the tool 

When enterprise IT teams get stalled at this adoption stage, it’s not always clear why. It may appear as though employees aren’t using it enough, or that they don’t understand all the capabilities, or that the model is a mismatch for current business challenges. But the root of the issue is that when IT stays in reactive mode, there’s still a human in every loop. 

Autonomy driven by AI can’t work to its potential and create any value until that loop is redesigned. IT teams have to move from a system of record to a system of action in order to get that AI value. That requires moving work out of the human queue entirely, rather than trying to make the queue faster for IT technicians. AI that simply makes suggestions can slow down IT workers further.

Here’s the difference between automation and true autonomy

What the teams that absorb AI do differently 

While lots of businesses have gotten stuck on AI acquisition and some on adoption, there are those who are trailblazing new ways of working with AI. A McKinsey study found that high performers use AI to drive growth, innovation, and saved costs. While 80% of respondents said their company’s goal for AI use is efficiency, the ones actually seeing value from AI are those that set growth or innovation as goals. And half of AI high performers plan to use AI to transform businesses, including redesigning workflows.

These takeaways are based on what we’ve learned about teams that have absorbed AI to take true advantage of what it can offer. Successful teams rewire their daily operations around these five areas:

1. Identify where resolution is autonomous

It’s common to apply new AI technology to solve problems that are deemed interesting — often complex or high-visibility engineering problems. However, teams that absorb AI focus instead on problems where resolution can be done completely autonomously, handled from beginning to end without the need for human intervention. 

Those might not be the most high-profile problems, but they are the ones that will drive change. For IT teams, a solid place to start is with high-volume, consistent ticket types, like password resets, software provisioning, and access requests. Instead of AI providing suggestions on fixing routine problems, teams that have absorbed AI can actually automate these common fixes. When subjective human judgment isn’t needed, and with guardrails in place, IT technicians can free up significant operational capacity with AI.

2. Define guardrails and approval boundaries

Without guardrails, approval boundaries, and other established guidelines, AI won’t gain the enterprise trust needed. So teams succeeding with AI are setting these guardrails up front, allowing AI to act instead of just advise. Otherwise, humans have to grant permission for any AI action, adding time rather than removing it.  

Teams that have absorbed AI have embedded policies into their infrastructure, with definitions around what AI is allowed to change autonomously, such as clearing a local cache or following an approval pathway for software access. The policies also include what AI can’t do, where a human gatekeeper is needed, such as modifying active firewall rules or changing schemas inside a production database. Modern AI tools should include audit trails, which provide a full accounting of what the system looked like before the intervention, what exact action (a script or call) the AI took, and what the system looked like afterward. IT technicians can review and remediate if needed.

3. Update your metrics

The metrics that have worked for many years are not the metrics that will take IT teams and organizations forward. Tickets deflected, such as by a chatbot, isn’t the right metric, because autonomous, AI-driven tools can actually remove tickets from IT’s queue. 

Consider more appropriate metrics like tickets eliminated and capacity reclaimed. These take into account the factor that a human is not doing all the resolving. In addition, metrics like system throughput and exception rate (where AI sends a ticket to the technicians) can be useful to understand how well the team has absorbed AI technology.

See why tickets deflected isn’t the right metric anymore

4. Redeploy capacity strategically

When autonomous AI is working as it should, you’ll see reclaimed capacity — up to 40%, such as with Robin by Atera. If you’ve only adopted AI, not absorbed it, this saved capacity can just represent a nonspecific, passive efficiency gain.  

The better next step is to be deliberate and strategic in redeploying all that saved capacity. IT teams that have absorbed AI use this reclaimed capacity as a valuable resource to apply to existing challenges. A service desk team can put that 40% of saved time into important initiatives that they haven’t had time for, like addressing technical debt, improving security, building a more resilient architecture, and more, all of which helps an IT team become proactive rather than reactive.

5. Choose context-aware AI

Not all AI technology is created equal, and there are different ways to acquire or adopt AI. For example, it’s possible to replicate legacy IT automation workflows with AI, which requires configuration by hand and manual updates for any updates.

Context-aware AI systems can learn your specific environment, including the knowledge base, commonly used scripts, and how the team typically resolves issues. Then, these AI platforms actively continue to learn, continuously parsing ticket resolution data, documentation, runbooks, and more. Rather than an engineer mapping every step of a particular workflow, context-aware AI can take in the intent of a request, then reference available information on how it’s been resolved before. With guardrails in place, the AI platform can solve the request autonomously, so there’s very little maintenance required.

Compare the best enterprise AI platforms for IT management

Can your organization absorb autonomy? An executive readiness check 

Using AI well, incorporating it deeply into the enterprise environment, is entirely possible. See these six areas to gauge to understand whether your organization is adopting AI or actually absorbing it.

1. Queue design

• Ask: Have we redesigned our work intake queues to account for the speed and volume of automated AI generation?
• Ideal absorption scenario: AI absorption requires queue design that optimizes for high-velocity, automated outputs, rather than legacy, human-paced workflows.

2. Guardrail clarity

• Ask: Are our risk, legal, and compliance boundaries translated into clear, real-time programmatic rules instead of dense text policies?
• Ideal absorption scenario: True AI absorption includes clarity on guardrails, where compliance boundaries are embedded directly into software code, not buried in static guidelines.

3. Metric model

• Ask: Are we measuring business-outcome velocity and system-wide throughput rather than old legacy metrics like individual employee hours logged?
• Ideal absorption scenario: Shifting to an AI operating model requires metrics that are centered on outcome velocity and system efficiency instead of on human labor hours.

4. Talent plan

• Ask: Does our workforce strategy focus on retraining employees for systemic oversight, quality assurance, and orchestration rather than manual execution?
• Ideal absorption scenario: An effective AI talent plan supports workers in moving from operational doers to cognitive orchestrators, focusing human talent on validation and edge cases.

5. Platform context-fit

• Ask: Does the AI tool fit naturally into end user environments, or do they have to constantly switch platforms or tools?
• Ideal absorption scenario: A high platform context fit means that AI capabilities live natively inside existing employee workspaces, instead of making employees use separate applications. 

6. Governance and audit

• Ask: Do we have continuous, automated tracking for model behavior, bias, and data lineage instead of relying on manual quarterly reviews?
• Ideal absorption scenario: AI governance and audits shift from manual reviews at one point in time to a continuous process of automated data tracking and real-time algorithmic logging.

As you get into product exploration, look for a native agentic platform with context from remote monitoring and management (RMM) and the help desk platform. These tools tend to be absorbed better than stitched-together solutions with an AI layer, as they can access specific business and IT data and continue learning without a lot of integration work.

Bringing it together with Atera 

It can be easier to move from acquisition to absorption with the right AI product. Robin by Atera serves as an autonomous layer: it resolves user issues autonomously right on devices, it learns your environment, escalates issues with full context to IT technicians, operates inside guardrails that you configure, and creates audit trails for every action.

The results of this type of AI lead to impressive outcomes: Robin by Atera achieves a 92% autonomous resolution rate, with a two-minute average resolution vs. 188 minutes when resolved by humans only. And Robin users eliminate or redirect 40% of their IT workloads, saving 11 to 13 hours saved per technician, per week. That’s across 13,000 customers in more than 120 countries, supporting 6 million devices. Robin guarantees initial results in less than 72 hours, and landed G2’s number one spot for AIOps in 2025, along with a Visionary spot in Gartner’s Magic Quadrant in 2026.

See how Robin can autonomously resolve your first tickets — with a guarantee of 50% of your Tier 1 and complex Tier 2 tickets in 90 days — and what your teams can do with the time back. Try it for yourself today.