Most AI initiatives I see in mid-market business-to-business companies are still being run as side projects. A CEO commits, and departmental leaders are told they own AI within their domains. Yet, 18 months later, the slides describing the program look almost identical to the ones that announced it—it’s still an aspirational work in progress, not a working reality.
The investment went in, but the business results and value didn’t follow. Something about how the business is run day-to-day keeps breaking the link between spend and outcome. It’s what I call the “bolt-on trap,” and it results in AI being layered on top of existing processes rather than rethinking the process from the ground up.
When investment is up, but value doesn’t follow
The data makes the gap hard to ignore. BCG’s Build for the Future 2025 study, which surveyed more than 1,250 firms globally, found that only 5% of companies are generating substantial value from AI. Another 60% report minimal returns despite real investment.
WRITER’s 2026 Enterprise AI Adoption Survey polled 1,200 C-suite executives and found that 75% admit their company’s AI strategy is more for show than actual internal guidance. This means that three out of four leaders are signing off on plans they themselves don’t believe will guide the business.
The market has also started signaling what serious commitment looks like. In 2022, only 2% of chief transformation officers held the role on a fully dedicated basis. A chief transformation officer is a senior executive dedicated solely to leading a major organizational change program. Essentially, acting as the CEO of the transformation itself, redesigning how the business runs across functions, with cross-functional authority, and direct CEO sponsorship.
By 2025, that figure had climbed to 48%, according to Monitor Deloitte’s 2025 Chief Transformation Officer Study. Companies serious about transformation have stopped treating AI transformation as a part-time job, but most mid-market AI programs haven’t made that move yet.
Anatomy of the bolt-on trap
In my work with mid-market B2B CEOs, the companies pulling ahead have all made the same structural move. A senior leader was given enterprise-level authority and dedicated time to lead the AI redesign across functions.
The companies stuck in the trap have an org chart with AI as an extra reporting line tacked onto existing jobs. The technology across both groups looks similar, but the structure around it tells the real story.
Or it works like this: Four or five executives carry AI accountability on top of full-time roles. No one has the calendar room to lead a cross-functional initiative or the authority to redesign anything beyond their own function.
Deloitte’s data suggests the market has begun to correct. Programs that work are increasingly run by leaders pulled out of their day jobs. Over half of the transformation teams in the 2025 study had members assigned full-time to the work.
That’s a meaningful break from how transformation has historically been staffed, and the reason is structural. In my experience, time is the input most correlated with transformation success, and bolt-on accountability can’t deliver it. Companies making real progress on AI have structurally separated AI leadership from day-to-day operational roles. That separation creates the available time to put into actual change.
For years, the AI conversation has been dominated by which model, which platform, and which AI tools. The organizational structure question has barely surfaced, and that’s where the real value gap lives.
Microsoft’s 2025 Work Trend Index argues that organizations cap AI’s value when they layer AI agents onto legacy roles, processes, and reporting lines. Forward-looking companies redesign around what the technology now makes possible, in how work gets done. AI layered onto existing work caps out fast.
3 Monday–morning questions
Here’s something to think about on your next Monday morning in the office:
1. Can someone in your organization name the three highest-value AI use cases?
2. Is there anyone who owns the cross-departmental prioritization?
3. Does that person have the bandwidth to actually run them?
When the honest answer is “no one full-time,” the next move sits in the organizational design, before any tool evaluation.
That move doesn’t have to mean hiring a chief AI officer or creating a new function from scratch. In most mid-market companies, a senior leader can be given the authority to make cross-functional calls. The leader can be assigned a small team that reports to them rather than to their old reporting line. The new structure has to back the mandate, or the mandate stays decorative.
The bolt-on trap is a design problem with a design fix. Companies that skip the structural work and lead with tool selection are showing up with substantial spend and minimal return. Acting on the technology before acting on the structure has been a reliable way to burn budget without changing how the business actually runs.
Debra Andrews is the founder and CEO of Marketri.