LaunchDarkly tops USD $200 million ARR as AI demand grows

LaunchDarkly has surpassed USD $200 million in annual recurring revenue, with revenue up more than 25% year on year.

The announcement coincided with the appointment of Andy Pemberton as Chief Revenue Officer as LaunchDarkly expands its global sales operation amid demand from large customers for tools to manage software and AI systems in production.

Known for its feature flagging software, LaunchDarkly said customers are increasingly spending on what it calls runtime control as businesses adopt AI in software development and operations. It argues that the challenge for engineering teams is shifting from generating code to managing the behaviour of autonomous systems after deployment.

That shift is central to LaunchDarkly's recent product push. Since introducing AgentControl in May, its AI-related pipeline has nearly doubled, with demand split evenly between new and existing customers.

AgentControl extends the company's existing infrastructure into AI operations. LaunchDarkly said its platform already handles more than 50 trillion feature flag evaluations each day and now lets teams adjust prompts, models, tools and policies without redeploying code.

Co-Founder and Chief Executive Officer Edith Harbaugh linked the revenue milestone to the company's original focus on balancing speed with oversight.

"When I co-founded LaunchDarkly, the core belief was simple: speed and control shouldn't be in tension. AI has made that more urgent than ever," Harbaugh said. "Teams aren't just shipping code, they're building systems of agents that act and evolve in production around the clock. LaunchDarkly provides the runtime control layer that makes these factories safe."

LaunchDarkly argues that software teams can no longer rely solely on pre-release checks when AI agents are involved because these systems continue making decisions after they go live.

"You can review a change before it ships; you can't review an agent's next thousand decisions," said Jonathan Nolen, Senior Vice President of Product at LaunchDarkly. "An agent doesn't misbehave in code review - it misbehaves in production, in front of your customers, so that's where control has to live. AgentControl puts it there: teams see what their agents are doing, and when something drifts they can correct it in seconds - no pull request, no redeploy, no waiting. That's what runtime control provides."

Leadership change

Pemberton joins from OutSystems, where he held a series of commercial and customer roles over nearly a decade, including Chief Customer Officer and Chief Revenue Officer. He will oversee sales, solution engineering, professional services, partners and alliances, sales operations and customer success.

The appointment points to a broader push into enterprise sales at a time when software suppliers are trying to define their role in AI adoption beyond model access and development tools. LaunchDarkly is positioning itself around the operational controls that govern how software changes are introduced and how AI-driven systems behave once they are running.

Harbaugh pointed to Pemberton's background in scaling global revenue teams.

"Andy has done this before; he's helped customers navigate transformations at scale, and that's exactly what we need for what comes next," Harbaugh said.

LaunchDarkly said it serves thousands of customers worldwide, including a quarter of the Fortune 500. Its latest figures indicate that established corporate users remain an important part of its growth, with existing and new customers contributing equally to the recent rise in AI-related sales activity.

The revenue milestone also places the company among a group of privately held software vendors that have moved beyond product-market fit into a more mature commercial phase, where investors and customers look for sustained growth, leadership depth and evidence that new product lines can generate demand.

For LaunchDarkly, that new product line is closely tied to how companies manage AI in live environments. Rather than focusing only on building models or generating code, it is making the case that businesses need tools to adjust and govern the behaviour of AI systems continuously once they are in front of users.

The company argues that the rise of AI agents shifts software control from deployment to production, making runtime decision-making a commercial category in its own right. Since the launch of AgentControl, its AI-related pipeline has nearly doubled, with demand split evenly between new and existing customers.