Chapter 4
90% of enterprise leaders describe themselves as confident in their AI agent readiness. Their rollback rate is the same as those less confident. The visibility data in this chapter shows why.
Across every role and every level of seniority, leaders are looking at the same AI challenges and seeing different things. When they don't share the same picture of what's happening, what's failing, and what's at risk, governance failures go unnoticed and investment goes to the wrong place.
Technical leaders (IT, engineering, security, and data – across all seniority levels) report rollbacks at 78%. Business leaders (finance, marketing, CX, product, and operations) at the same organizations report them at 69%. The gap extends to the nature of the incidents: technical leaders report PII-related issues at 34% while business leaders report them at 27%. Business leaders are less likely to know a rollback happened, and they’re also less likely to know why.
The same pattern holds for the engineering burden. Technical leaders report their teams spending more than half their time on guardrails at a rate 9 points higher than business leaders (38% vs 29%). The engineering cost is real, but it’s not landing equally across the leadership team.
And yet technical leaders are more confident than their business counterparts. Technical leaders describe themselves as very confident in their AI readiness at 54%, compared to 43% of business leaders. Having a line of sight into how the AI program is being built and run – even when it surfaces more failures – appears to positively impact confidence compared to having no technical visibility at all.
of technical leaders have shut down or rolled back a live AI agent. (Sinch, 2026)
of business leaders have shut down or rolled back a live AI agent. (Sinch, 2026)
of technical leaders are very confident in their AI readiness. (Sinch, 2026)
of business leaders are very confident in their AI readiness. (Sinch, 2026)
Across the C-suite, confidence in AI readiness is generally high: 60% of leaders describe themselves as “very confident” in their organization’s AI readiness. But two leaders sit significantly below that. Only 45% of CPOs and 38% of CISOs define themselves as “very confident” – and what they have in common is proximity to what failure actually costs. They understand the risk and feel it most when it happens.
For CPOs, every governance failure lands directly on the product roadmap. Every rollback is a sprint derailed, a timeline attacked, a revenue opportunity delayed. That exposure produces a different relationship with confidence.
For CISOs, it’s something more specific. Customer data exposure is the leading cause of AI rollbacks at 44%, and the CISO is the person who has to explain it when it happens. The closer you are to understanding what goes wrong, the harder it is to be confident it won’t.
Sinch data (2026) shows only 45% of CPOs and 38% of CISOs define themselves as “very confident.”
The C-suite’s confidence is high. But the C-suite isn’t the one rebuilding the agent after a rollback. They’re not triaging the support queue when the bot goes down or explaining to a customer why their data surfaced in a conversation it shouldn’t have. The people doing that work assess their AI readiness very differently.
60% of C-suite leaders describe themselves as very confident in their organization’s AI readiness. That number drops to 49% at VP level, and 43% among Directors and Managers. The pattern holds regardless of industry, AI maturity, region, or organization size.
That 17-point gap between the C-suite and the people running the program highlights a difference in exposure. And it runs one way: the further from the consequences, the more confident the assessment.
of C-suite leaders describe themselves as very confident in their organization’s AI readiness. (Sinch, 2026)
of VPs describe themselves as very confident in their organization’s AI readiness. (Sinch, 2026)
of Directors and Managers describe themselves as very confident in their organization’s AI readiness. (Sinch, 2026)
Vertical spotlight
The more senior the leader, the further they are from what happens when AI fails – the engineering rebuild, the support surge, the customer trust recovery. That distance shapes what they believe about how the program is performing, across every vertical.
FIVE LEADERS, ONE AI PROGRAM, FIVE DIFFERENT VIEWS
CIOs, CTOs, IT and engineering leads have the clearest view of what’s actually in production: 65% report active deployments, the highest of any group, likely because they own the infrastructure it runs on. But visibility doesn’t mean immunity from failure. These leaders report rollbacks at 76% – 2 points above average.
FIVE LEADERS, ONE AI PROGRAM, FIVE DIFFERENT VIEWS
CPOs and product leaders invest the most in trust and security at 81% yet report the most rollbacks (82%) and the lowest guardrail maturity at 29%. Only 3% say their current provider isn’t falling short. Part of that may be these leaders’ definition of failure – broader than any other persona because the product roadmap sits at the intersection of every governance issue, every integration failure, and every sprint derailed by safety infrastructure.
FIVE LEADERS, ONE AI PROGRAM, FIVE DIFFERENT VIEWS
Of all the roles in this study, the CISO and security leaders understand the problem best. That hasn’t made it easier to solve, though. They report fully mature guardrails more than any other group at 38% – but that still leaves six in ten not fully there. 41% spend most of their time on safety and still score lower than any other on satisfaction with their provider’s security and compliance infrastructure. The people with the clearest view of the problem are the least satisfied with the tools available to solve it.
FIVE LEADERS, ONE AI PROGRAM, FIVE DIFFERENT VIEWS
To deliver AI-powered customer experiences, CMOs, CXOs, and other marketing and CX leads depend entirely on infrastructure they don’t own. And that dependency shows: They are the least likely to define themselves as very confident at 45%, and report one of the lowest deployment success rates. For a CMO, the success of an AI deployment is directly tied to its impact on customer experience, which explains why – when an AI agent fails – CSAT damage is their primary concern.
FIVE LEADERS, ONE AI PROGRAM, FIVE DIFFERENT VIEWS
The CFO organization controls the budget but not the infrastructure. Their primary concern when AI agent fails is brand damage – which, for a financial leader, translates directly to financial exposure. That framing extends to how they think about risk before a failure even happens: At 53%, they are the most worried about data leakage of any group – 7 points above average and 15 points above the CISOs and security leaders responsible for preventing it. And yet finance leaders remain the most confident group in the study at 53% very confident.
When leaders experience different realities of the same incident, the time it takes to respond to failure increases. It’s during that time that irreversible brand damage happens, where incomplete information leads to misdirected investment decisions, and where the next failure gets funded.
The organizations most exposed are the ones where confidence increases the further you get from the failures – where the sprint log says one thing, the board update says another, and no one with budget authority has seen both.
The visibility gap is not a reporting problem. Adding another dashboard won’t solve it. The gap is structural, and it’s built into how different roles experience the same AI program: their goals, their involvement, how they measure success differently, and how they define failure.
Two questions worth asking before your next AI deployment: What’s the biggest risk in your current program? And where’s the biggest opportunity to move faster?
If your CFO and your CIO give different answers, the gap is already shaping what gets fixed and what gets funded.
