The Best AI Transformations Don't Come From Consultants or IT. They Come From Your Own People.

Here is a number that should stop every executive in their tracks: 42% of companies scrapped most of their AI initiatives in 2025, up from just 17% the year before. (S&P Global Market Intelligence)

Billions of dollars. Dozens of pilots. Consultants engaged. Software purchased. And nearly half of all enterprise AI efforts abandoned before they reached production.

The question worth asking is not why did AI fail? The question is who was driving it?

Because the data is increasingly pointing in one direction: the organizations quietly winning with AI are not the ones who hired the most outside experts or handed the mandate to IT. They are the ones who put the people closest to the work in the driver's seat. This approach has a name. It is called citizen-led AI adoption.

Citizen-led AI adoption is not about giving every employee access to ChatGPT and hoping for the best. That is uncontrolled experimentation — and it carries real governance and security risk.

Citizen-led adoption is a structured strategy in which business teams — finance, operations, HR, accounting, revenue cycle, customer service — are actively equipped and empowered to identify AI use cases, redesign their own workflows, and implement solutions with appropriate governance guardrails in place.

It is the deliberate transfer of AI capability into the hands of the people who understand how work actually happens.

That distinction matters, because most organizations are currently doing the opposite.

1. Outsourcing to Third-Party Consultants.

The AI consulting market hit $3.75 billion in 2024, nearly tripling from the prior year. (National CIO Review) Demand is not the issue. Results are. The core problem is structural: when strategy is separated from operational implementation, failure follows. Research from B-Works found that 80% of consulting-driven transformations fail when strategy is disconnected from execution.

Consultants, however talented, do not work your close process. They do not sit in your AP queue. They build a solution, hand it over, and leave. What they leave behind is often a system nobody fully understands and capability that lives entirely outside the organization. The moment the engagement ends, so does the momentum.

2. Hiring More IT Talent Without Operational Context.

The second instinct is to hire more technical staff and task them with identifying AI opportunities across the business. The problem is not the talent. It is the lens. IT teams are exceptional at governance, security architecture, systems integration, and infrastructure. But identifying where AI creates operational value requires deep, daily familiarity with how business processes actually work.

The finance team knows which reconciliations take three days when they should take three hours. Operations knows where approval workflows stall. HR knows which onboarding steps cost them new hires. Gartner confirmed this in 2024, noting that individual contributors — the people responsible for the majority of automatable tasks — are often the most underserved when it comes to AI support and guidance.

3. Unstructured Shadow AI.

The third failure mode is already happening inside your organization right now. Microsoft research found that 70% of organizations struggle to equip their workforce with AI skills, yet employees are not waiting for permission. The result is shadow AI: unsanctioned tools, unreviewed outputs, and proprietary data entering public interfaces. Shadow AI is citizen adoption without the governance. It is not the answer — but it is a clear signal that your workforce wants to solve problems with AI and is not getting the structure to do it safely.

The research on what separates successful AI organizations from failing ones points consistently toward one theme: the depth of capability built inside the business, not the sophistication of tools deployed.

When business teams are structurally equipped to lead AI adoption — with the right training, governance framework, and operational methodology — four things happen that outsourcing and IT-only approaches simply cannot replicate.

Citizen-led adoption is not a strategy for bypassing IT. It is a strategy for partnering with IT differently. Technology teams remain essential — they own security architecture, system integration, enterprise enablement, and data infrastructure. The governance frameworks that protect your organization from regulatory exposure require technical leadership.

What changes is the division of responsibility. IT is a critical partner in enabling citizen-led adoption — not the sole driver of it. The most effective AI organizations in 2025 have stopped asking their technology teams to identify business value and started asking them to help business teams implement it safely.

Citizen-led adoption done well begins with the people in the workflow mapping what actually happens — not what the org chart says should happen — and identifying where friction, manual effort, and rework are creating the most cost. Only then does the conversation about AI tooling become productive.

That sequence matters. Process first. Technology second. People always.

If you are an executive thinking, we do not have the internal structure to execute this — that is the right diagnosis. Most organizations do not, yet. The practical starting point is an honest assessment of your organization's current AI readiness: where your teams stand on AI literacy, where your highest-value operational opportunities exist, and what governance infrastructure needs to be in place before you scale.