What your CEO must know about AI (AI column)
I saw something interesting. speech Hubert Mistel, Senior AI Researcher at Novartis Hubert leads a team of artificial intelligence researchers in drug development and also teaches machine learning and data science at Dublin Business School. The event was held at the AI Engineer World's Fair. 2025 The largest technical AI event of the year that has passed 3-5 June 2025 year in San Francisco. Below are the key ideas in my interpretation.
1. The Agency Future of Organizations Hubert began with a provocative statement: in three years, organizations can be run by AI agents. This follows from the pace of technology development. He goes on to talk about what an AI agent is, an application based on large language models that has autonomous behavior. The key feature of agents is that they perform step-by-step actions with dynamic control, adapting the use of tools depending on the results of previous steps.
2. Transforming workflows Agential workflows are the main difference from traditional automation. While earlier machine learning could automate individual steps, AI agents can: Automate and delegate several steps at the same time Act as a link between tasks Work without human intervention, requesting feedback only when necessary
Next, the author reflects on the archetypes of employees, among which he highlights the following: Silent performers – low communication, high productivity Individual performers – balanced profile Links - connect different teams Multipliers – strengthen the work of other team members Knowledge centers - provide expertise
3. The New Reality of the Labor Market Domain knowledge and intelligence are becoming cheap and easily accessible, which means that Companies and employees should strive for multidisciplinary knowledge. n New roles will appear: “miners of work processes”, “human orchestrators of AI” AI accelerates “actors” – the difference in performance can reach not 10x, but 100x (Sounds like a reference to 10x engineers.)
4. Democratizing the creation of agents It is critical to allow employees to create agents using no-code/low-code tools. Cognitive self-awareness – understanding one’s own thought processes Ability to translate mental steps into specific instructions Access to tools to quickly create agents
5. Ethical issues There are two critical issues that need to be addressed to further spread the agents: Are we ready to not just delegate tasks, but allow AI agents to manage processes?
- What values to put in agents and how to resolve conflicts between humans and AI?
6. Practical recommendations for managers Strategy Audit: If your AI strategy remains relevant when you replace “AI agent” with “ML” – it is outdated Workflow mapping: Detailed understanding of workflows is critical for maximum benefit Readiness for transformation: the organizational structure of the company can change dramatically Investing in context: context becomes the new oil, more important than data
All in all, Hubert had an interesting talk about the AI-agent revolution, which has already begun and is worth getting ready for.
#AI #ML #Software #Engineering #Development #Agents #Architecture