At last week’s IE Executive MBA 2025 Round Table, I joined Bernardo Crespo and Lucia Egea to speak on the theme: In Search of the Technological El Dorado – Entrepreneurship, Innovation, and Surviving the Whirlwind of AI Change. It was a lively and thoughtful session, but what made it particularly meaningful was the calibre of questions from Lucia and the EMBA students—each incisive, each rooted in real executive uncertainty.
What follows is not a transcript, but a structured reflection on six questions that deserve serious attention in boardrooms and strategy meetings across every industry. Each one touches on the profound shifts being driven by AI—from generative models to autonomous agents. And each one gets to the heart of the dilemma: how should we lead when the technology is faster than our institutions?
1. What Kind of Jobs Will AI Really Transform?
The adoption of generative AI has been the fastest in technological history. What began with LLMs now leads us to agents—self-directed systems operating across digital workflows. This isn’t a theoretical shift. Protocols for agent interoperability (A2A, MCP, etc.) are emerging rapidly, and with them a new kind of operating model.
To understand which jobs are at risk, consider how enterprises have evolved over the past three decades. We moved from desktop tools like Excel to prescriptive ERP systems. Many roles emerged to bridge applications: management accountants summarizing data, merchandisers calculating reorders across spreadsheets. These roles, built around mechanical data movement and structured logic, are now low-hanging fruit for agents.
But not every job is so easily reduced. Roles that require trust, persuasion, contextual judgment—these demand the nuance of human intelligence. This is why consultants, especially at large firms with leverage-heavy models, are exposed. Their pyramids rely on junior staff executing repetitive research and process work. A well-trained internal team, using AI tools, can now do much of the same. Consultancies will need to evolve, or risk irrelevance.
2. Why Is AI Outpacing Enterprise IT?
AI’s development cycle is now faster than most enterprise release plans. Consumers can access state-of-the-art tools for free. Employees bypass security policies to use ChatGPT through personal accounts. Meanwhile, vendors release new models on a weekly basis, making it nearly impossible for CIOs to lock in a stable tech stack.
There’s a paradox here. On one hand, CTOs face pressure to act—to deploy GenAI or agentic systems before competitors gain a strategic edge. On the other hand, the technology is still maturing. The agentic stack is not yet enterprise-grade in many respects: security, governance, and auditability are not where they need to be for regulated environments.
So how should an executive respond? If AI provides a clear competitive advantage now, then it’s worth investing in a robust and reliable in-house solution. But if the use case is not mission-critical, the more prudent choice may be to wait. The pace of development is so rapid that enterprise-ready tools will soon be available at lower cost with better features. The key is to understand where your organization sits on that spectrum.
3. Why Is Agentic AI a Game-Changer for Traditional Industries?
Most of our enterprise processes were designed around human limitations. Take the grocery supply chain: a merchandiser uses tools to forecast demand, plan inventory, and place orders. Because a human can only track a certain amount of data and make a limited number of decisions, we organize around categories and assign responsibilities accordingly.
But agents don’t have those constraints. Agentic AI can analyze massive, interdependent data sets—sales, logistics, promotions, external signals—in real time. Agents can forecast hourly. They can negotiate directly with supplier agents to optimize outcomes across the whole value chain.
This isn’t just efficiency. It’s a redefinition of the business model. If your competitor has deployed such agents and you haven’t, they’ll price better, stock better, and satisfy customers faster. As always in retail: the margins are thin, and the smarter operator wins.
4. Will AI Create a Generation That Can’t Think?
There’s a fair concern among educators and employers: are we raising graduates who let the AI do all the thinking?
My view is the opposite. If we teach the fundamentals well—critical thinking, logic, and ethics—then AI becomes a powerful amplifier of those skills. Today’s MBAs are learning not just to memorize frameworks but to orchestrate systems of tools and agents to get results. It’s a higher-order competency.
Just as we don’t worry that calculators made accountants stupid, we shouldn’t fear that GenAI makes strategists soft. The issue is fluency: are your people comfortable collaborating with AI? If not, you’re training them for obsolescence. Because whether you plan for it or not, every device, every app, every job will soon have AI embedded.
5. Can We Trust the Agents?
AI does not “think” in a human sense. It mirrors patterns. This means it also mirrors our biases. One well-known case: Amazon’s recruiting AI downgraded female candidates—not because it was sexist, but because historical hiring patterns taught it that successful engineers were men. The lesson? If our processes are biased, AI will scale that bias unless we actively correct it.
Agents present a different challenge: identification and intent. In the near future, your website won’t just serve human users—it will serve agents too. These agents won’t need images or calls to action. They’ll be crawling for facts, reviews, and structured data.
But not all agents are equal. In fact, 90% of them may be malicious—bots spamming systems or scraping content for fraud. A further 9% will be benign crawlers. Only a fraction will represent legitimate customer value. The question becomes: how do you know which is which?
Technologies like digital IDs, verifiable credentials, and personal data wallets will help distinguish trustworthy agents from the noise. Just as we taught our systems to detect spam and fraud in email, we must now do the same for agent interactions.
6. How Should an Organization Introduce GenAI and Agents?
This is the human part of the equation—and it’s covered in Agents Unleashed Vol 1. Business psychologist Aparna Uberoy laid out a seven-step plan that’s both practical and emotionally intelligent:
- Anticipate external influence: Don’t wait for rumors—lead with facts through internal communications.
- Be transparent early: Let people know what AI is being introduced for—and what it’s not.
- Involve employees as collaborators: Workshops, surveys, and listening sessions help reduce resistance.
- Redesign roles, don’t eliminate them: Focus on how AI augments, not replaces.
- Co-design the tools: User involvement in development builds better tools and better adoption.
- Pilot with purpose: Small wins build momentum and uncover hidden issues.
- Train comprehensively: Upskilling should be framed as a career advantage, not a survival tactic.
This isn’t just change management. It’s about building psychological safety around a new era of work.
Conclusion: Keep Your Humanity in the Loop
AI agents are evolving fast. Faster than policy, faster than product roadmaps, faster than most companies can absorb. But no matter how sophisticated these systems become, they are still tools—ours to design, direct, and discipline.
The future belongs not to the most technical, but to the most intentional. Stay curious. Stay adaptive. And above all, stay human. Let your agents be brilliant, but let your humanity shine through them.
If your organization is preparing to build an agentic strategy, I invite you to explore Agents Unleashed—and to reach out. This moment demands more than understanding. It demands action.