How IT Services Can Scale AI Beyond Pilots: The Operating-Model Shift

IT services companies scale AI by changing how work is delivered—not by adding more tools or pilots. The practical shift runs from helping individuals use AI, to redesigning end-to-end workflows, to reshaping roles and the operating model around new ways of working. Each stage needs clear accountability, workforce plans and measures of business value.

Why AI pilots often fail to become operating results

A pilot can show that a model performs a task. It does not, by itself, show that the task can be integrated into a live service, that its output meets quality requirements, or that customers and employees will adopt the new process. Scaling means resolving those issues across the workflow, with ownership for decisions, exceptions and outcomes.

#	Preview	Product	Price
1		MINISFORUM MS-02 Ultra Workstation Mini PC, Intel Core Ultra 9 285HX (24C/24T, up to 5.5GHz), PCIe...	$1,659.00	Buy on Amazon
2		GMKtec EVO-X2 AI Mini PC Ryzen Al Max+ 395 Superchip 128GB LPDDR5X 2TB SSD	$3,649.99	Buy on Amazon

As an Amazon Associate I earn from qualifying purchases.

The gap is visible in several studies, though their figures should not be treated as one combined estimate. McKinsey & Company reported in March 2026 that nearly two-thirds of organizations had not scaled AI beyond a few pilots, and no more than one in ten reported AI-agent use beyond pilot stage within a business function. In a separate study conducted in the second quarter of 2025, Boston Consulting Group (BCG) surveyed 1,250 companies worldwide: 5% generated measurable value from AI at scale, 60% reported no material value, and 35% were scaling efforts and seeing some returns. BCG defines measurable value as increases in revenue and cash flow, and improvements in processes and workflows.

The figures use different surveys, dates and definitions. Together they point to a delivery problem, not proof that any particular tool or organizational design will succeed.

#1 Best Overall

MINISFORUM MS-02 Ultra Workstation Mini PC, Intel Core Ultra 9 285HX (24C/24T, up to 5.5GHz), PCIe 5.0 x16, 32GB RAM 1TB SSD,USB4 v2 80Gbps, Dual 25GbE+10GbE+2.5GbE, Wi-Fi 7, 350W PSU

• High-Performance AI Processor：The MS-02 Ultra features an Intel Core Ultra 9 285HX (24C/24T, up to 5.5 GHz, 13 TOPS NPU), delivering fast and efficient performance for AI inference, algorithm development, and media workloads. A PCIe x16 expansion slot supports desktop-class GPU upgrades for advanced model training and accelerated computing tasks. It's ideal for creators, engineers, and teams handling intensive parallel workloads.
• 4 × M.2 PCIe 4.0 + 4 × DDR5 SODIMM slots:Four DDR5 SODIMM slots support up to 256 GB of memory, while ECC helps maintain data integrity in mission-critical environments. Four PCIe 4.0 M.2 slots support up to 24 TB of storage, supporting RAID 0/1/5/10, combining high-speed performance with data protection. It allows for the creation of independent scratch disks, media libraries, and project drives, providing high-throughput for production workflows.
• PCIe & USB 4.0 v2: Up to three PCIe slots can be equipped, including a dual-slot x16 GPU. The main slot supports PCIe 5.0, meeting the needs of high-bandwidth creative and computing workloads. USB 4.0 v2 (80Gbps) supports high-bandwidth external storage and displays.
• Ultra-fast Networking: Wi-Fi 7 further enhances wireless performance with next-generation speeds and low-latency stability. Intelligent bandwidth switching optimizes throughput in different network environments, ensuring optimal performance for enterprise or local networks. Dual 25GbE ports (providing up to approximately 3.125 GB/s bandwidth, about 25 times faster than traditional 1GbE), enabling seamless large-scale file transfers and parallel computing. 10GbE and 2.5GbE ports, with support for Intel vPro technology, ensure enterprise-grade remote management and deployment flexibility.
• Server-grade thermal architecture: Utilizing a dedicated CPU/GPU airflow design, equipped with a 6-pipe dual-fan cooler, it maintains stable performance even under sustained loads, delivering up to 140W Turbo power while maintaining a 100W TDP, and operating with noise levels as low as 36 dB. An integrated 350W power supply ensures stable and reliable output for demanding computing tasks and fully loaded extended configurations.

$1,659.00

Buy on Amazon

Three horizons: from individual assistance to operating-model reinvention

McKinsey’s September 2026 analysis describes three horizons for AI adoption. They are useful as a way to identify the actual scope of a company’s effort—not as a mandatory maturity ladder or a guarantee that every organization should pursue the same design.

1. Enablement: help people use AI in their jobs

At this horizon, employees get access to general-purpose tools and guidance for individual tasks. The work may become faster or easier, but workflows, decision rights and service responsibilities largely remain as they were. Access and usage are starting points; they are not evidence that a service line has captured value.

2. Automation: improve cross-functional workflows

At this horizon, teams connect AI to a workflow that crosses roles or functions—for example, the steps involved in handling a service request. Teams need to map handoffs, define what information the system can use, set quality thresholds and decide when a person must review or take over. The target is a better workflow, not simply an AI feature embedded in one task.

3. Reinvention: redesign roles and how the service operates

Reinvention changes the broader design of work: who does which tasks, how expertise is deployed, how exceptions are handled, and how the service is governed and measured. The change can affect role boundaries, staffing and the operating model, so it requires leadership decisions and workforce planning as well as technology.

In McKinsey’s 2026 operating-model survey, 48% of respondents classified in the reinvention horizon reported meaningful enterprise value from AI, compared with 24% in automation and 13% in enablement. The survey included 701 English-speaking executives and senior leaders across regions; the reinvention subgroup contained 53 respondents. McKinsey cautions that these panel responses are individual perspectives, not representative accounts of the respondents’ organizations. The results show an association in respondents’ reports, not proof that reinvention caused the difference or that a specific company will achieve it.

Choose an operating model around the work

There is no single operating model established as best for every IT services company. A useful design choice is not simply “centralized or decentralized”; it is deciding which decisions need common standards and which are best made close to a service or customer. Compare the options against the work you intend to change.

These are design axes, not a scored ranking. A company can centralize platforms, security and baseline controls while leaving workflow choices and service-specific measures with accountable business teams. The right division depends on the workflow, risk and expertise required; the sources do not establish a universal allocation of authority.

Redesign the workflow before automating it

Start with a service process that has a clear owner and a real business problem. Map its steps, handoffs, information inputs and exceptions before deciding what AI should do. Automating a fragmented process can make its weaknesses harder to see and its mistakes faster to propagate.

1. Define the outcome. State what should improve for the customer or the business, such as workflow performance, service quality, cost or growth. Record the current baseline and how the measure will be collected.
2. Map the work. Identify repeatable, data-heavy tasks and distinguish them from tasks that require judgment, customer context or accountable decisions. Include exception paths and handoffs, not just the routine case.
3. Assign ownership. Name the business or service owner, the people responsible for the model and data, and the person accountable for accepting or escalating output in the workflow.
4. Set review and escalation rules. Define which outputs need human validation, what quality threshold is acceptable for the task, and what happens when information is incomplete, uncertain or outside the workflow’s scope.
5. Integrate and monitor. Put the capability into the process employees already use, then track both operational quality and the intended business outcome. Review exceptions and failures as part of normal service management.
6. Expand only when the workflow works. Use the evidence from the live process to decide whether to refine it, extend it to another team, or redesign roles and responsibilities more broadly.

This sequence keeps a technology demonstration separate from a service change. A pilot can still be useful, but its success criteria should test the workflow, controls and outcome that the company intends to scale.

Make workforce change part of the design

AI adoption changes the mix of work people do, even when job titles do not change immediately. Planning only for tool training leaves a gap between knowing how to use a system and knowing how responsibilities, quality expectations and career paths will evolve.

The World Economic Forum’s January 2026 article, authored by HCLTech CEO and Managing Director C. Vijayakumar, recommends a shared skills taxonomy, learning pathways connected to role changes, and internal mobility. It also argues for dividing repeatable, data-heavy work from work requiring human judgment and accountability. These are recommendations from an IT services executive’s perspective, rather than independent evidence that a particular workforce program produces a defined outcome.

For a service organization, workforce planning should answer practical questions:

Rank #2

GMKtec EVO-X2 AI Mini PC Ryzen Al Max+ 395 Superchip 128GB LPDDR5X 2TB SSD

• EVOLUTION RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
• AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
• AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
• EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
• QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.

$3,649.99

Buy on Amazon

• Which tasks are likely to change, and which still need a person to make or own the decision?
• Which existing roles gain new responsibilities, and which capabilities must employees learn to perform them?
• How will employees move into changed or emerging roles rather than being left with a tool and unclear expectations?
• Who is responsible for maintaining skills as workflows and systems change?

McKinsey’s 2026 operating-model analysis also identifies leadership modeling, role clarity and continuous learning as organizational practices associated with AI value. That finding is an association, not a causal guarantee. Leadership matters in practice because teams need visible priorities and decisions about how work changes—not just encouragement to experiment.

Build governance into the workflow

Governance is most useful when it answers what an employee or system should do at the point of decision. A general policy is not a substitute for workflow-specific instructions about review, exceptions and accountability.

McKinsey’s November 2025 State of AI report says higher AI performers are more likely to define when model outputs require human validation. It also describes six dimensions for value capture: strategy, talent, operating model, technology, data, and adoption and scaling. The reported relationship between practices and value should be read as correlation, not proof that adopting any one practice will cause higher returns.

• Human validation: Specify which outputs require review and who has authority to approve, correct or reject them.
• Quality thresholds: Set task-specific expectations for accuracy and completeness, with a defined response when those expectations are not met.
• Bias and risk checks: Determine what checks are relevant to the workflow and how their results affect deployment or continued use.
• Escalation: Give employees a clear route for unusual cases, uncertain outputs and service-impacting failures.
• Accountability: Identify who owns the service outcome and who can change the workflow or pause its use.

Controls should match the consequences of an error. A low-impact drafting aid and an AI-supported decision in a customer-critical workflow should not be assumed to need identical review.

Measure value beyond access and activity

Tool availability, logins and prompt counts can show whether people are using a system. They do not establish that AI has improved the service. Choose measures that connect the workflow change to its purpose, and track quality alongside efficiency so a faster process does not conceal worse outcomes.

• Cost: Track the cost of delivering the defined service or workflow, using a consistent baseline.
• Growth: Where relevant, measure whether the change contributes to revenue or another explicitly defined growth outcome.
• Service quality: Monitor customer-facing quality and errors alongside speed or volume.
• Workflow performance: Measure the process outcome the redesign is intended to change, including exceptions or rework where relevant.

BCG’s Q2 2025 study illustrates why definitions matter: its reported “measurable value” category includes revenue and cash-flow increases as well as process and workflow improvements. BCG also reported that the technology function’s share of AI value rose from 7% in its 2024 study to 13% in its 2025 study. Those percentages describe BCG’s studies and its measure of the technology function’s share of value; they are not a forecast for an individual IT services provider.

Where services partners fit

A services partner can contribute delivery capacity, domain expertise or an AI-enabled managed service. The meaningful distinction is whether it is simply helping implement tools or taking responsibility for changing how a defined service is delivered. The latter requires agreement on workflow scope, quality, escalation, accountability and outcome measures.

Accenture’s March 2025 discussion of IT operating models presents AI-enabled managed services as a way to automate workloads, deliver faster and address knowledge gaps. That is a provider perspective and strategic proposition, not independent proof of outcomes for any particular provider. The World Economic Forum’s 2026 article likewise argues for moving from isolated services and tools toward IP-led solutions scaled through technology and industry ecosystems. These directions may inform a service strategy, but they do not establish that every company should build or buy a particular model.

For IT services leaders, the practical test is whether the partner’s role is clear: what part of the workflow it owns, how it shares knowledge with the client, who remains accountable for decisions, and how both parties will assess the service outcome.

What the wider workforce forecasts do—and do not—say

The World Economic Forum’s January 2026 article reports that around 1.1 billion jobs could be transformed by technology over the next decade and that AI and information processing will affect 86% of businesses by 2030, citing the Future of Jobs Report 2025. These are broad forecasts reported in the article, not estimates of job losses or predictions for IT services companies specifically. They reinforce the scale of workforce planning as a strategic issue, but do not identify which roles at a given company will change or how quickly.

A practical test for moving beyond pilots

Before scaling an AI initiative across a service line, leaders should be able to answer these questions with specific owners and evidence:

• Is there a named workflow owner and a defined business or customer outcome?
• Has the team redesigned the full workflow, including handoffs and exceptions, rather than automating one isolated task?
• Are role changes, skills development and internal mobility part of the plan?
• Are human review, quality expectations, escalation paths and accountability clear to the people doing the work?
• Can the company measure service outcomes and quality against a baseline, rather than relying on usage figures alone?
• Is the division of responsibilities between central teams, service teams and any external partner explicit?

If those answers are missing, the next step is usually to clarify ownership and workflow design, not to multiply pilots. If they are in place, the company can use live operational evidence to decide whether to expand, adapt or stop the initiative.

Quick Recap

Bestseller No. 1

MINISFORUM MS-02 Ultra Workstation Mini PC, Intel Core Ultra 9 285HX (24C/24T, up to 5.5GHz), PCIe 5.0 x16, 32GB RAM 1TB SSD,USB4 v2 80Gbps, Dual 25GbE+10GbE+2.5GbE, Wi-Fi 7, 350W PSU

$1,659.00

Bestseller No. 2

GMKtec EVO-X2 AI Mini PC Ryzen Al Max+ 395 Superchip 128GB LPDDR5X 2TB SSD

$3,649.99

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.