Ethan Mollick put it plainly in a recent post: “A corporate position that workers should ‘just use AI to do stuff’ has never been enough. AI use in companies is a leadership problem.” That framing names something many mid-market companies are now sitting in the middle of.
The tools are deployed. The licenses are paid. The all-hands memo went out. And 80% of the organization is using AI inconsistently, superficially, or not at all.
Closing that gap is what a fractional CAIO is actually hired to do.
sequenceDiagram participant CTO as CTO participant BU as Business Unit Leaders participant CAIO as Fractional CAIO Note over BU: AI tools deployed CTO-->>BU: Integration complete BU-->>CTO: Low adoption CAIO->>BU: Department use-case workshops CAIO->>BU: Workflow redesign with AI CAIO->>BU: Measurement and accountability Note over BU: Consistent adoption
Why the CTO Usually Can’t Fix This
The CTO’s job is to build and maintain the technical systems the business runs on. That includes AI infrastructure: model selection, API integrations, security posture, data pipelines. A strong CTO handles all of that.
What a CTO typically cannot do — and shouldn’t be asked to do — is change how the operations team runs its workflows, how the finance team builds its reports, or how the customer success team handles its caseload. The CTO doesn’t run those teams and doesn’t have the organizational mandate to redesign how they work.
AI adoption is a workflow redesign problem. Every business unit that genuinely uses AI has changed how it works — what happens in the morning, who reviews what before it goes out, which tasks are now handled by an AI before a human touches them. That change doesn’t happen from a technology mandate. It happens through someone at the executive level working with each team to understand their work and identify where AI changes it.
That’s the CAIO’s job.
What the Adoption Gap Actually Looks Like
The pattern is consistent across mid-market companies in every industry. The engineering team uses AI heavily — for code generation, code review, documentation, and increasingly for agentic workflows. One or two enthusiastic individuals in other departments run their own experiments. Everyone else uses AI the same way they use Google: occasionally, for search-adjacent tasks, without it affecting the fundamental structure of their work.
The gap is not a skills problem, and it’s not a tool problem. The tools are capable enough. The gap is a problem of translation — nobody has helped the operations manager understand what their team’s actual work looks like when AI is in the loop, or helped the finance director understand which of their 40-step reporting processes can be compressed to eight.
I saw an analogous pattern during a reporting initiative at a large health insurer — a #169 Fortune 500 company with $6 billion in revenue and complex data across dozens of business units. The organization had a deep and capable reporting platform, and when it launched, most business units continued producing their reports the way they always had. The system existed; the adoption lagged. Closing that gap required going department by department, working with each team’s subject matter experts to understand what they actually needed and demonstrating how the new system changed their work. It wasn’t a technology problem. It was a translation and ownership problem.
The same dynamic plays out in AI adoption, at higher velocity.
What a Fractional CAIO Does About It
The adoption work a CAIO does is unglamorous and essential. It breaks into three parts.
Use-case discovery, by business unit. This is not a survey or a workshop where people list ideas. It’s the CAIO sitting with department heads, understanding what their teams actually do, and identifying the specific places AI changes the output of that work. Most business units have two or three high-leverage use cases. Finding them requires knowing the work and knowing the tools — which is why it requires an executive who can operate at both levels.
Workflow redesign and rollout. Once the use case is identified, someone has to define the new workflow — what the AI-assisted version of that process looks like, what the human review steps are, what quality looks like, and what changes in how the team is organized around the work. This is change management, not technology deployment. A CAIO who can run this process at the department level is doing the work that separates organizations with 20% adoption from those with 80%.
Measurement and accountability. Adoption without measurement isn’t progress — it’s activity. The CAIO owns the metrics that tell leadership whether the AI program is producing the business results it was supposed to produce: time saved, output quality, error rates, cost per unit of work. Those metrics don’t exist by default. They have to be defined and tracked, and the CAIO is the person accountable for them.
The Organizational Position That Makes This Possible
A CAIO who reports to the CTO can close the engineering adoption gap. They cannot close the adoption gap across business units that don’t report to the CTO.
The organizational structure that works is a CAIO reporting to the CEO or COO, with a mandate to work across every function. That position gives them the standing to work with the CFO on finance AI workflows, the CMO on marketing AI, and the COO on operations AI — without having to route through the technology org for permission to engage.
PwC’s 2026 CAIO survey found that 76% of CEOs have hired or plan to hire a Chief AI Officer, up from 26% two years earlier. The organizations moving fastest on that number aren’t doing it because AI infrastructure is complex — most companies have a CTO who can handle that. They’re doing it because AI adoption across a non-technical workforce requires a different kind of executive ownership than technology deployment does.
The fractional model makes that executive level accessible to mid-market companies that can’t justify the $400,000 to $650,000 base salary a full-time CAIO commands. The work is the same; the cost structure is not.
