Replay: AI doesn’t need better models; it needs braver leaders. Putting humans at the helm of enterprise AI.

Enterprise AI is already making decisions inside most large organisations, and almost none of them can tell you who owns those decisions when something goes wrong. That was the starting point for this HFS webinar, where Phil Fersht and Dana Daher were joined by Paul Daugherty, former Accenture CTO, now AI Advisory Chair at TPG, and Raj Sundaresan, CEO at Altimetrik, to walk through findings from the new Humans at the Helm of AI study of 500+ senior leaders across Global 2000 companies and five industries.

The conversation moved past the “human in the loop” slogan to something harder: who actually owns the outcome when the machine gets it wrong, and why so few enterprises have answered that question.

Watch the replay + download the presentation

• Watch the webinar replay
• Download the presentation deck

Key takeaways

• Individuals are racing ahead of their employers on AI, creating what Phil calls the “AI velocity gap”: people experiment with frontier tools on their own time, then return to siloed systems and outdated workflows at work.
• Data sharply separates high and low AI maturity: 88% of high-maturity firms report double-digit CX gains, 41% report double-digit new revenue, and just 3% lack clear AI ownership.
• Only 40% of leaders have a formal AI strategy—and 15% have none at all. As Raj noted, the gap isn’t a lack of interest, but a lack of clarity on who owns AI and what that role entails.
• AI is a participant sport for the C-suite. As Paul argued, CEOs and boards must actively use AI themselves rather than delegate fluency to the CIO.
• “Human in the loop” is often becoming audit theatre, with reviewers rubber-stamping outputs without clear thresholds, real authority, or traceability into decisions.
• The impact of AI is playing out at the task level, not the job level. Despite predictions of widespread job loss, roughly 80% of companies are using AI to drive growth rather than reduce headcount.
• The most underrated workforce skill is the ability to challenge AI outputs. Yet 46% of leaders already worry about over-reliance, and 80% of organisations provide fewer than 10 hours of AI training per year.
• A Chief AI Officer can become a convenient scapegoat if AI isn’t embedded across the business. Raj advocates a federated, seven-layer approach spanning data, governance, observability, and model retraining.
• Services partnerships must shift from billing for effort to sharing accountability for outcomes. Raj describes a “diamond” delivery model—heavy hitters supported by forward-deployed engineers—replacing the traditional pyramid.
• 2026 may be the year a small group of enterprises (10–15%) achieve landmark AI transformations, forcing entire industries to respond while much of the Global 2000 risks falling behind.

Ready to move from AI pilots to real accountability?

If you’re trying to work out who should actually own AI decisions in your business, close the velocity gap between your people and your processes, or rethink your operating model so humans are genuinely at the helm, HFS can help. We work with enterprise leaders to get past the experimentation loop and build AI strategies that hold up when the decisions start getting real.

Contact us to explore how to apply these insights to your AI and enterprise transformation journey.