What Top AI Labs Are Really Doing With Observability

Predictions and Expectations:

• Olivier expects the industry to face "a very, very different deal" where public investors have less concentration than VCs, making it "really, really hard to understand what things look like three, four, five years from now."
• Olivier believes that with AI changing rapidly, companies "can't wait to see the demand materialize from the customer base so you need to get ahead of it."
• Olivier expects that because AI changes "so fast," companies like Datadog "are going to get wrong more often" and "need to be at ease with that."
• Olivier predicts that because the "process looks like a year from now" is unknown, teams will not know "how many PMs, how many designers, how many security people, how engineers you need to build a thing."
• Olivier notes that smaller AI labs with "infinite compute" might use tools in ways "the rest of the word is not going to do," as most companies will not be "happy to you know incinerate you know a billion in compute."
• Olivier states that he is "not sure I found the right profile yet" for roles like Head of Sales and advises to "find someone who looks good, hire them right away" rather than waiting.
• The speaker expresses that they "need to be better" and "we're gonna do it so for example you know, we're going to do differently we're gonna be better."

Timelines and Milestones:

• A co-founder predicted "in two quarters, we're not writing any code anymore," describing an inversion where developers "mostly automate, sometimes write" instead of the reverse.
• The speaker notes that the way you write code "has changed three time in the last year."
• The speaker states they had a "moment in in December" when "the models got really good, the threshold was crossed."
• The speaker mentions they have been public "since 2019" and that their lockup expired "the day of the COVID lockdowns."
• The speaker notes that public investors "come in and out of these names a couple of times a year as opposed to a vc that's going to be in like seven eight companies and and you know stick around for five to ten years."

Technology and Product Direction:

• Datadog plans to "invest a lot so that the product can be used by agents."
• Datadog is "building a lot the smarts directly into the product" so that "the system can do everything on its own."
• The speaker expects to "pivot the organization" to adapt to AI, acknowledging "it's going to be painful it is painful but we're doing it."
• The speaker plans to "take more shots and be wrong more often" due to the speed of AI changes.
• The speaker describes a shift from "mostly write, sometimes automate" to "mostly automate, sometimes write."

Market and Industry Outlook:

• The speaker expects it is now "absolutely critical to go after the U.S. market as soon as you can" after achieving "some form of market fit."
• The speaker believes "you can get funded pretty much anywhere in Europe" now, unlike when he started in 2012.
• The speaker notes that "public investors have less it's for them it's harder to reason about about things" compared to VCs.
• The speaker predicts that the market has "a ton of disruption" and that for public investors, "it's really, really hard to understand what things look like three, four, five years from now."
• The speaker observes that the "AI labs, like the top two or three AI labs in the world" use the product in "weird ways" due to "infinite compute."

Company Plans:

• The company is "pivoting the organization" to handle the shift where "smaller teams can do a lot more," allowing exploration with "a team two or three" instead of eight.
• The speaker plans to "get ahead of" AI demands rather than waiting for customer materialization.
• The company is "building a lot other things that we see them [customers] do" based on usage of their extensions and workflows.
• The company plans to "get ahead" on strategic projects where "nobody's asking us to do that, but it looks like the world is going that way."

Financial Guidance:

• The speaker mentions a past drop from "$10 billion to being worth, I don' know, $4 billion or something like that" during the lockup expiration, noting the market went "up like crazy again" afterwards.
• The speaker notes that if a decision is "unpopular with the market," the company might "lose no 20 30 40 percent" in stock value, which "creates a comp issue" regarding RSUs.
• The speaker states that the "main risk is not the survival of the company it's kind of like losing key performers yeah because their issues are worthless."

Risks and Caveats:

• The speaker admits that strategic bets where "nobody's asking us to do that" mean "you tend to be wrong often because you just made that up."
• The speaker warns that in the AI era, "things are changing so fast that you just can't wait to see the demand materialize."
• The speaker notes that relying on "infinite compute" labs might provide a "glimpse of what's coming up after" that is "not necessarily the most representative" because "your most companies are not going to be very happy to you know incinerate you know a billion in compute."
• The speaker states that the destination of AI adoption is unknown, as "nobody knows what the process looks like a year from now."
• The speaker acknowledges that "firing someone in your company, you feel horrible" but notes the rest of the company often thinks, "What took you so long?"