At First American — the world’s largest title insurance company at the time, with 900 engineers across 15 subsidiaries, 770 applications, and a database containing 100 million US property records — my role as enterprise architect included evaluating acquisition targets. In 2009 and 2010, that meant reading the code of companies the business was considering buying. When I recommended walking away from a roughly $100 million acquisition after finding that the technical reality did not match the deal thesis, the recommendation carried weight because the accountability was clear. A technology executive had evaluated the target and was willing to stand behind the conclusion.
That accountability model has not changed in the AI era. What has changed is the set of decisions the technology executive needs to make.
journey title CEO's Clarity Through the Fractional CTO Engagement section Orientation Technology audit complete: 2: CEO Risk surface visible: 3: CEO section Delivery Vendor decisions accelerate: 4: CEO AI governance installed: 4: CEO section Steady State Board reporting cadence: 5: CEO Accountability model clear: 5: CEO
What Actually Changed in the AI Era
The narrative around AI and technology leadership tends toward two directions: AI is making technology decisions easier and reducing the need for executive oversight, or AI is creating new categories of risk that require more oversight. Both are partially true. The more accurate frame is that AI has shifted which decisions require executive attention, not the total amount of attention required.
The decisions that have become easier: many routine technology choices, vendor evaluations for well-understood categories, and day-to-day development execution. AI tools have made developers significantly more productive, which reduces the executive bandwidth required for delivery oversight.
The decisions that have become harder, or newly necessary: evaluating AI vendor claims, governing AI-generated code at scale, managing a workforce where AI augmentation is unevenly distributed across the team, assessing whether AI capability claims in M&A targets reflect technical reality or positioning, and owning the accountability framework for when AI-assisted processes fail. These are executive decisions. They do not get easier because AI tools are involved — in several cases they get harder because the technology is less legible to non-technical stakeholders.
The Vendor Evaluation Problem
One of the clearest examples of what the fractional CTO needs to own in the AI era is AI vendor evaluation. The market has produced a large number of vendors claiming AI capabilities that vary significantly in their technical substance. A vendor claiming an “AI-powered” platform may have purpose-built models trained on proprietary data with documented performance benchmarks — or it may have a wrapper over a general-purpose model API that any developer could replicate.
Both describe themselves as AI-powered in marketing materials. The difference is material to anyone signing a multi-year platform contract. Evaluating which is which requires technical depth — code-level and architecture-level review of the actual implementation, not just the product demo.
This is the same evaluation I performed at First American. The acquisition target presented well. The data room was clean. The capability claims were specific. A documentation-level review would have moved to confirmation. Direct access to the code and database — which I requested and received — surfaced the actual technical reality. The recommendation to walk away from the roughly $100 million deal, before the check cleared, was the product of that direct access. That is still the evaluation methodology. The domain has changed; the approach has not.
The Governance Accountability
The second major ownership area in the AI era is governance accountability. When an enterprise with hundreds of developers using AI coding tools produces a security incident traceable to AI-generated code that was not properly reviewed, the accountability chain runs through the technology executive who was responsible for the governance framework.
This is not hypothetical. Enterprise vibe coding governance is an active gap at most mid-market companies right now — the tools are in use, the review standards have not caught up, and the accountability model for what happens when something goes wrong is undefined. The fractional CTO who establishes that framework before the incident is doing executive-level risk management. The company without that framework is accumulating risk without the accountability structure to manage it.
The Accountability That Does Not Come with the Subscription
The broader point is this: AI tools are subscriptions. A fractional CTO is an accountable executive. When the AI tool generates code that fails in production, the tool vendor’s support team will not be in the post-incident review explaining what went wrong and how to prevent it next time. The technology executive will be.
That accountability function — being responsible for outcomes, not just inputs — is what the fractional model provides. One to three days per week of a technology executive who owns the decisions and is accountable for the results, rather than a collection of tools that can be turned off if they stop working.
The AI era has made software development faster, more accessible, and in some respects more complex to govern. None of those changes removed the need for an accountable technology leader. They shifted what that leader needs to focus on, and in a few areas they raised the stakes for getting it right.
