Episode Details

Denis O'Shea is the founder and CEO of a technology services company, and we spoke about how a painful sales failure became the foundation for 22 years of helping organizations extract more value from technology. Early in his career at Nokia, a customer challenged him on why they should buy more technology when employees barely used what they already had. Denis says that conversation “burnt a piece of my brain.” He later built a 250-person mentoring operation that worked one-on-one with a million people across multiple countries, and today applies those lessons to AI, security, and workplace productivity.

His own company’s AI rollout exposed exactly why enthusiasm is not enough. After deploying AI to roughly two-thirds of the organization, the team discovered 33,000 sensitive files that were overexposed, gaps in employee training, no clear foundation for AI agents, poorly defined use cases, and no objective way to demonstrate ROI. Their response became a five-part method: define use cases, secure and classify data, train people, establish a secure foundation for agents, and measure the economic value of AI-supported work.

Denis also explains how extreme strategic focus changed his company’s trajectory. From New Zealand, his team committed to becoming exceptionally good at one narrow technology specialization, eventually winning a global partner award and gaining introductions to major enterprise customers. The discipline, he says, was to “say no to 99 things” while continuing to say yes to one thing for years. The same philosophy now informs his view of AI: build security in from the beginning, prepare for potentially hundreds of agents per employee, and expect companies to face three growing management problems—data, agents, and spend.
For listeners, the practical value is a concrete framework for adopting AI without losing control of security, costs, focus, or measurable business outcomes.

Key takeaways

• Define AI use cases before deciding who receives the technology.
• Audit and classify sensitive data before exposing it to AI.
• Train employees beyond browser-based AI into everyday productivity tools.
• Give every AI agent clear ownership, permissions, policies, and lifecycle management.
• Measure AI ROI at the task level, not through adoption alone.
• Say no to 99 opportunities to become exceptional at one.