One of our core beliefs as a firm is that AI has, and will continue to, significantly eat into labor TAM, leading to a proliferation of agents capable of performing work historically done by humans. These agents need to be reliable, successfully interacting with core business processes and the external world to accomplish these tasks, and long-running.
We believe Temporal has become the de facto solution to enable these requirements; we’re thrilled to announce Lightspeed’s investment in Temporal’s $300 million Series D. Temporal is the Durable Execution platform that many of the world’s most demanding AI systems already run on — and it’s becoming one of the most important pieces of infrastructure in the agentic era.
The Pattern We Kept Seeing
Over the past year, we spoke with hundreds of teams building AI-powered products. The conversations almost always followed the same arc: building a demo is easier than ever, but bringing the vision into production is hard.
Not because the models weren’t smart enough. Because the systems around them couldn’t handle real-world execution.
Teams building legal workflows, coding assistants, and enterprise automations kept hitting the same wall. They would start with homegrown job processing and state management systems. However, as they sought to scale their user base, have their agents take multiple steps and make intermediate tool calls, or run agents for minutes rather than seconds, their systems would fail. The problem would compound over longer running horizons because errors rates compound. Retries would duplicate work. State would get lost mid-execution. Engineers would get paged to manually restart processes that were supposed to be autonomous.
Temporal enables teams to go from a cool demo that works some of the time to an experience that can recover, is auditable and ultimately works consistently in production.
Why Temporal Wins
Temporal solves this by making Durable Execution feel like writing ordinary application code. The underlying concept — recording every step so a process can pick up exactly where it left off after a failure — has existed for years. But previously, developers had to work directly with event stores and schemas, which was complex enough that most teams couldn’t justify the effort.
Temporal’s platform abstracts away this complexity. Developers write normal code. Temporal wraps it so execution resumes correctly after failures, without developers ever having to think about events, retries, or state recovery. It just works.
The problem is universal across any software system, from business processes to infrastructure management to payments to now AI applications. The company first took flight with the rise of distributed systems and microservices architecture, but is now on ‘steroids’ with the rise of AI.
Temporal’s co-founders, Samar Abbas and Max Fateev, have spent over two decades working on this problem. They built five versions of this system at Amazon, Microsoft, and Uber before spinning out to build Temporal, the culmination of all their efforts and learnings.
Temporal’s Traction
Temporal is experiencing tremendous adoption. The company saw over 380% y/y growth and is exceeding 20M installs per month. The company has garnered adoption from major AI labs, such as OpenAI, to leading AI agent companies, such as Replit and Lovable, to traditional Fortune 500 companies, such as Yum! Brands and ADP.
Because Temporal can be applied to virtually any workflow, its potential use cases are quite vast. A development team may first adopt Temporal to power their AI agents, but then expand usage to orchestrate microservices, handle payment workflows, process data, and manage their infrastructure. Temporal ultimately represents a fundamental shift in the way code should be written and executed.
Our Partnership
Lightspeed has had a long history of partnering with highly technical infrastructure founders, and we are excited to continue this tradition with Samar and Max. We believe we are moving to a world where every worker may interact with multiple agents per day, each of which must durably complete hundreds of ‘actions’ and potentially run for days, months, or years at a time. But for that to happen, those agents need to be reliable, recoverable, and long-running. They need infrastructure that was built for exactly this kind of problem.
Temporal was built to orchestrate distributed systems from the outset, and the rise of agents, the ultimate distributed system, has made the problem it solves impossible to ignore. We’re thrilled to partner with Samar, Max, and the entire Temporal team as they define what production-grade AI infrastructure looks like.
Authors
