Datadog CEO Olivier Pomel on AI, Trust, and Observability

Introduction

In this episode of "AI Native Dev" brought to you by Tessl, host Guy Podjarny welcomes Olivier Pomel, CEO of Datadog. Olivier, a seasoned expert in the field of cloud computing and observability, co-founded Datadog in 2010 with Alexis Le-Quoc. Before founding Datadog, Olivier was instrumental in building data systems for K-12 education as the VP of Technology at Wireless Generation, where he expanded the development team significantly before the company's acquisition by News Corp. He also has a background in software engineering with experience at IBM Research and has contributed to the development of the VLC media player. Olivier holds a Master's degree in Computer Science from Ecole Centrale Paris. His extensive experience in both technical development and leadership roles positions him as a trusted voice in the intersection of AI and cloud technology, offering valuable insights into the evolution of observability and AI-powered applications.

Main Discussion Topics

Understanding Observability

Olivier describes observability as the modern evolution of monitoring. While the term itself might seem passive, it represents an active engagement in understanding system behaviors. As Olivier mentions, "Observability is the modern version of what used to be called monitoring before." Datadog's approach to observability includes anomaly detection through products like Watchdog, which monitors systems and alerts users to irregularities.

AI-Powered Applications and Audience

The conversation shifts to AI-powered applications, discussing how they are designed and who they serve. Olivier highlights the importance of understanding the application's audience and ensuring that AI integrations align with user needs and expectations. He notes, "These are applications built that are powered by AI," emphasizing the need for tailored solutions that resonate with specific user demographics.

Challenges in AI and Security

Olivier discusses the challenges faced in AI, particularly in security. He mentions how control planes and data planes merge in the context of large language models (LLMs), complicating traditional security approaches. Olivier states, "You can try and inject stuff through logs, through traces, through pretty much everything that submits data," highlighting potential vulnerabilities like prompt injection and the need for robust security measures.

Evolution of AI and User Interaction

The podcast delves into the evolution of AI interactions, from co-pilots to autonomous agents. Olivier shares insights on how these interactions can enhance user experience and productivity. He emphasizes, "There's two issues there. One is what's the right form factor for the customers to interact with it the right way," stressing the importance of building trust and ensuring low-friction adoption of AI products.

Future of Observability and AI

Olivier envisions the future of observability in AI, discussing how tools like LLM observability might evolve. He discusses the potential for personalized models for large customers and the importance of immediate value from AI products. Olivier expresses, "We see actually very good progress there and we can clearly see on the horizon the moment where the technology is good enough to be put into the hands of customers." The conversation also covers the integration of UX in AI systems and how it might transform over the next five years.

Building AI-Powered Teams

The discussion concludes with insights into staffing and organizational changes needed to support AI development. Olivier talks about the evolving profiles of engineers in AI development and how Datadog adapts to these changes to remain at the forefront of innovation. He notes, "It's critically important to have low friction for products to be adopted and to show value," underlining the need for adaptable and skilled teams.

Summary/Conclusion

This episode provides a comprehensive view of how observability has evolved with AI's rise, offering listeners insights into the challenges and opportunities that come with integrating AI into modern applications. Key takeaways include:

• Observability is an active process crucial for modern AI applications.
• Security remains a complex challenge in the AI landscape.
• Future AI systems will likely feature more autonomous interactions and personalized models.
• Building effective AI teams requires adapting to new engineering profiles.