Why Traditional Observability Falls Short for AI Agents - The Data Exchange with Ben Lorica Recap

Podcast: The Data Exchange with Ben Lorica

Published: 2026-01-22

Duration: 43 min

Summary

In this episode, Lior Gavish, CTO of Monte Carlo Data, discusses the shift in data teams from traditional analytics to AI-driven automation and the resulting need for enhanced observability for AI agents. He emphasizes that as these teams scale their AI initiatives, the complexities necessitate new tools and approaches to ensure reliability and trust.

What Happened

Lior Gavish, CTO and co-founder of Monte Carlo Data, joins Ben Lorica to explore the evolving landscape of data teams as they transition from traditional analytics to AI and automation. Lior highlights that when Monte Carlo Data started seven years ago, the focus was primarily on data quality and observability. However, as data teams have evolved into data and AI teams, the emphasis has shifted towards ensuring the trustworthiness of AI agents in production. This transition reflects a broader trend where automation is increasingly taking over tasks that were once handled manually by data engineers.

Lior explains that this shift has significant implications for how data teams operate. Previously, data engineering roles involved building and maintaining pipelines primarily for analytics. Now, many of those tasks are being automated by AI, freeing up teams to focus on more strategic initiatives. He notes that the entry-level data engineering job has transformed, with many basic tasks being automated through AI tools, allowing teams to scale their efforts and innovate at a faster pace. This change is not limited to tech companies; Lior emphasizes that industries across the board, including manufacturing and education, are adopting AI tools for various applications, thereby broadening the impact of this technological shift.

As organizations begin to deploy AI agents in production, the need for observability becomes paramount. Lior describes how Monte Carlo has adapted its solutions to provide observability specifically tailored for AI agents. This involves the introduction of observability agents that automate monitoring and troubleshooting workflows, addressing the growing complexity that comes with operating AI systems. He concludes by reflecting on the accelerating adoption of these technologies, suggesting that we are on the verge of an inflection point in the use of AI tools across different sectors and geographies.

Key Insights

Key Questions Answered

What is the evolution of data teams towards AI?

Lior explains that data teams have shifted significantly from focusing primarily on analytics to becoming data and AI teams. This evolution reflects a broader trend in which organizations are increasingly building agents to automate various business functions. The initial focus of Monte Carlo Data was on data quality and observability, but as the industry has changed, the emphasis has moved towards ensuring that AI agents are reliable and trusted in production environments.

How has AI changed the role of data engineers?

The role of data engineers has transformed dramatically as AI technologies have begun to automate many of the tasks that were once performed manually. Lior notes that entry-level data engineering jobs have shifted from building pipelines to leveraging AI tools for automation. This transition allows teams to focus more on innovation and strategic initiatives rather than routine grunt work.

What challenges do organizations face with AI adoption?

Organizations encounter a variety of challenges when adopting AI, particularly regarding scalability and complexity. Lior highlights that while many companies are beginning to deploy AI agents in production, the overall percentage of organizations that have fully scaled these initiatives remains relatively low. However, he notes that there is a growing number of companies that are preparing to go live with AI agents, indicating an impending shift in the landscape.

How important is observability for AI agents?

Observability has become a critical need for organizations as they deploy AI agents. Lior discusses how the complexity of operating these systems necessitates new monitoring tools. Monte Carlo Data has introduced observability agents to automate monitoring and troubleshooting workflows, addressing the challenges teams face in managing and ensuring the reliability of AI systems.

Is AI adoption limited to tech companies?

Lior confidently asserts that AI adoption is not confined to tech companies. He mentions that their customer base spans various sectors, including manufacturing, media, and education. AI tools are being utilized across different industries for a wide array of applications, suggesting that the trend of integrating AI into workflows is becoming ubiquitous.