No PriorsNo Priors Ep. 21 | With Datadog Co-founder/CEO Olivier Pomel
At a glance
WHAT IT’S REALLY ABOUT
Datadog CEO on AI, observability, security, and disciplined hypergrowth strategy
- Olivier Pomel, Datadog’s co-founder and CEO, traces the company’s origins from bridging dev–ops culture gaps to becoming a unified observability and security platform at massive scale. He explains how building from New York, staying close to customer reality, and designing a single integrated platform underpins Datadog’s broad product expansion and efficient, near-profitable growth. A significant portion of the discussion focuses on generative AI: its impact on software workloads, developer productivity, observability, and the emerging LLM tooling stack, as well as Datadog’s cautious, outcome‑driven use of AI in its products. Pomel also details Datadog’s approach to acquisitions, security, customer segmentation, and leadership practices that sustain execution through changing macro and technological environments.
IDEAS WORTH REMEMBERING
5 ideasDev–ops collaboration, not metrics, was Datadog’s original core problem to solve.
Datadog began as a way to get development and operations teams seeing the same reality and working together, and only later evolved into the full observability platform most people recognize today.
Operating from New York forced capital efficiency and closer alignment with real customer needs.
Skepticism from Bay Area investors and a smaller local deep-tech pool led Datadog to run near-profitable from early on, focus obsessively on product–market fit, and benefit from higher employee retention versus the Bay Area.
A unified platform is Datadog’s main strategic moat but requires heavy ongoing investment.
Roughly half the company works on the core platform, and every acquisition is re-platformed in year one, which is costly but critical to delivering deeply integrated, end‑to‑end workflows across many product areas.
Generative AI shifts value from writing code to understanding, operating, and securing it.
As developers become far more productive and write more software they understand less deeply, demand grows for tools that help observe, debug, secure, and manage increasingly complex, AI‑augmented systems—exactly the layer Datadog serves.
LLMs open new observability use cases but don’t replace precise numerical methods.
Datadog still relies on classical statistical and numerical models for anomaly detection and alerting, while using LLMs to combine heterogeneous data (metrics, logs, state, docs) into richer insights and explanations where fuzziness is acceptable.
WORDS WORTH SAVING
5 quotesThe starting point for Datadog was not monitoring or even the cloud; it was, “Let’s get dev and ops on the same page.”
— Olivier Pomel
If one person is ten times more productive, they’ll write ten times more stuff—but they’ll understand what they write ten times less.
— Olivier Pomel
Everybody’s buying security software. Nobody is more secure as a result.
— Olivier Pomel
There are great medicines today for security, but for them to work you need to inject them in every single one of your organs every day. We want to deliver it to you in an IV.
— Olivier Pomel
With LLMs we clearly have ignition. We might have liftoff soon. The question is whether we need a second stage.
— Olivier Pomel
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome