Jun 15, 2023 · 44m · no-priors
No Priors Ep. 21 | With Datadog Co-founder/CEO Olivier Pomel
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of No Priors, Datadog co-founder and CEO Olivier Pomel discusses building a unified observability platform, navigating disciplined enterprise growth from New York City, and the profound impact of generative AI on software development and LLMOps.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 19.2% of the talking time here. How this is scored →
The hosts as informed peer 5.4 Guest teaching 5.1 Guest disagreement 1.4 The hosts pushing back 1.6
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 32:41 Dismantling customer tolerance for false positives
Olivier dismisses naive assumptions about AI alerting, bluntly explaining that despite what customers claim in surveys, two false alerts at night will cause them to turn off a product forever.
Hardest push from the hosts ▶ 34:29 Elad challenges the timeline on automated operationsElad pushes back on Olivier's conservative automation stance, demanding whether the obstacle is a fundamental technological ceiling or merely the early stage of enterprise implementation.
Biggest teaching moment ▶ 32:41 Why ChatGPT fails debugging without runtime stateOlivier provides a definitive masterclass on why LLMs give incorrect root-cause diagnoses on raw stack traces unless integrated directly with live program execution state and numerical telemetry.
The host holds their own ▶ 14:32 Elad articulates enterprise LLM routing architectureElad demonstrates sharp domain authority by explaining how advanced enterprises dynamically route prompts between frontier models and low-cost self-hosted open-source models.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Olivier Pomel's Background and Datadog's Founding Vision | 4 | 3 | 1 | 1 | Sarah sets a collaborative tone by highlighting her personal familiarity with Datadog's acquisition of Screen. Olivier shares the founding insight that Datadog was born from interpersonal dev and ops friction rather than monitoring or cloud architecture. | |
| The NYC Advantage and Origin of the Datadog Name | 4 | 4 | 1 | 1 | Sarah and Elad prompt Olivier on building an enterprise infrastructure company in NYC despite Bay Area skepticism. Olivier breaks down the strategic advantages of avoiding the Silicon Valley echo chamber and achieving superior engineer retention. | |
| Datadog Core Platform and the Generative AI Demand Surge | 5 | 5 | 1 | 1 | Elad frames Datadog's scale and asks how generative AI shifts observability needs. Olivier delivers an insightful breakdown explaining that AI productivity gains will shift enterprise value from writing code to understanding and operating it. | |
| The Rapidly Changing AI Stack and the Evolution of LLMOps | 6 | 6 | 2 | 2 | Elad notes rapid enterprise stack shifts and LLM routing dynamics. Olivier contrasts the fast-moving open-source GenAI wave with past multi-year transitions like Kubernetes, explaining why LLMOps will succeed where classical MLOps struggled. | |
| Unified Platform Architecture and Disciplined Acquisition Strategy | 6 | 5 | 1 | 1 | Elad and Sarah explore Datadog's unified platform architecture and M&A discipline. Olivier explains the rigorous strategy of spending the first year post-acquisition completely rebuilding tools onto the unified platform and requiring product shipment within three months. | |
| Product Expansion to APM and Ubiquitous Cloud Security | 6 | 6 | 2 | 2 | Sarah probes how Datadog can expand into security when conventional wisdom views CISO point solutions as separate from IT monitoring. Olivier reframes the problem, arguing that ubiquitous runtime deployment by developers delivers superior security outcomes compared to top-down sales tools. | |
| Combining LLMs with Runtime State and the Limits of Automation | 7 | 7 | 2 | 4 | Sarah and Elad drill into the reality of automated SRE and LLM hallucination in production. Olivier explains why raw LLMs fail stack trace analysis without runtime state, and Elad pushes back to clarify whether technology limits or organizational adoption block full operational automation. | |
| Disciplined Execution, Talent Identification, and Serving All Customer Segments | 5 | 5 | 1 | 1 | Sarah asks about managing multi-segment customer bases and identifying top talent during market shifts. Olivier outlines his view of star performers as black holes that absorb and resolve organizational complexity, and explains why modern cloud standards allow serving startups and Fortune 100 clients simultaneously. |
Statements from this episode (19)
Olivier Pomel was an original author of the VLC media player
“This led me later on to be one of the first authors of VLC, the media player who you know, I think is mostly used for viewing illegally downloaded videos. And I should say most of the people who made that successful came in after me, picked up the project and …”
Pomel: Datadog was founded for Dev-Ops collaboration, not monitoring
“So the starting point for Datadog was not monitoring it was not even the cloud initially. It was, let's get dev and ops on the same page. Let's give them a platform, someplace you can work together, see the world the same way.”
Pomel: Datadog hovered around profitability throughout almost its entire existence
“We built a company that was fairly efficient from day one and, you know, hovered around profitability throughout its whole existence pretty much.”
Pomel: NYC has less tech talent than Bay Area, but retention is much higher
“So it's a bit more difficult to recruit in New York. There's less pure tech talent. There's less deep tech talent in New York than there is in the Bay Area, but the retention is a lot higher. So, you know, if you give people, you know, great responsibilities, …”
Pomel: Chatbots may not be how users interact with AI in two years
“It's possible that some of the things we've seen with LLMs, you know, where, you know, all of a sudden everything's a chat bot. It's possible that it's not the way people want to interact with everything, you know, two years from now.”
Pomel: No Competitor Has Caught Up to OpenAI's Frontier Model Performance
“Nobody has quite caught up to open AI yet in terms of what the frontier model is and the maximum level of performance you can get.”
Pomel: Datadog Avoided MLOps Tooling Because Startups Struggled for Traction
“And that's a field, by the way, we're only watching from afar over the past few years. And the reason for that was that, you know, we saw a hundred different companies do that but few of them, you know, reaching true traction.”
Pomel: LLMs Expand AI Tooling Users From Data Scientists to All Developers
“And I think today it's changed quite a bit because LLMs are the killer app. Everybody is, is trying to use them. And the users, instead of just being a handful of data scientists in every company, end up being pretty much every single developer.”
Pomel: Datadog splits engineering evenly between platform and product use cases
“So the rule of thumb for us is about half the team is on the platform. And, but that relates to what we do, right? We sell a unified platform. So internally, you know, as I said, half is on the platform. The other half is more on the product side, like specifi…”
Pomel: Infrastructure, APM, and log monitoring will merge into one category
“An example of that being observability is emerging as one big super category that really encompasses what used to be infrastructure monitoring, application performance monitoring, and log management. And we still sell those three as different skews today. But …”
Pomel: Datadog spends the first post-acquisition year completely re-platforming technology
“When we do so, the first thing we do is we actually re-platform the companies we've acquired. So we spend the first year really post-acquisition Rebuilding everything that the company had built on top of a unified platform.”
Pomel: Datadog requires every acquired company to ship something within three months
“We have a plan basically that calls for shipping something together within three months of the acquisition, which is very short because the acquisition people celebrate a little bit, you know and then they have to get oriented and you HR systems and whatnot. T…”
Pomel: The main acquisition risk is demoralizing existing staff, not wasting money
“The main risk is that it's not that you waste the money with the acquisition, It's that you demoralize every, everybody else in the company because they see these new companies being acquired and they don't understand the value or they wonder, you know, why yo…”
Pomel: Datadog spent its first six or seven years on one product
“We spent the first six or seven years of the company on our first product.”
Pomel: Everyone buys security software but nobody is more secure
“Everybody's buying security software. Nobody is more secure as a result.”
Pomel: Two false positive alerts at night make users disable tools forever
“The reality is you send them two, two false positives at night the same week and they'll turn you off forever.”
Pomel: Real production use cases for LLMs will likely arrive within a year
“We might have liftoff probably in the next year or so with real production use cases, because right now most of the stuff is still not production. It's still demo ware and, you know, Private beta and that sort of stuff.”
Pomel: Following where work naturally flows reveals an organization's top performers
“The best people just obliterate problem areas. You, yeah, black holes for problems. You send problems, they disappear. And that's where, you know, you can promote them. You can easily find them in organizations because
[2484] Olivier Pomel: You see all the wor…”
Pomel: Shared cloud infrastructure allows software vendors to serve startups and enterprises alike
“In the cloud and open source generation, like the tooling is the same for all companies. If you go back 15 years, if you were a startup, you were building on open source. And if you were a large company, you were buying whatever Oracle or Microsoft was selling…”