Written by: Techub News Organized

Recently, the Y Combinator channel released an interview with Trigger.dev co-founders Matt Shillingford and Eric. Trigger.dev is a platform that helps developers integrate and run AI Agents and complex asynchronous workflows in their products. The interview reviewed the team's three-year journey since joining YC and revealed how their product evolved from the initial concept of a "developer version of Zapier" to the core infrastructure supporting a multitude of AI Agent applications today. This conversation is of great reference value for understanding the evolution of developer tools in the AI era, the practice of open-source business models, and how startups can seize opportunities presented by shifts in technological paradigms.

Product Evolution: From Background Job Framework to AI Agent Execution Platform

The story of Trigger.dev began with the YC Winter 2023 batch. Initially, their product idea was to become a "developer version of Zapier", aiming to provide developers with an easy-to-use automation tool connection framework. The team invested significant effort in product design and user experience, even featuring code snippets as a primary display content on their official website because they believed that developers care most about "what the code looks like." This precise understanding of developer psychology allowed the product to gain a decent response in early communities like Hacker News.

However, the market feedback on the first version indicated that this was not a point capable of sparking growth. They quickly realized that while there was demand for the initial "backend office" automation use cases (such as connecting GitHub and processing marketing data), the product's match to the market was not high. Thus, the team pivoted to the second version: an SDK focused on embedding asynchronous tasks into user products. This meant that developers could integrate long-running tasks (such as document processing and video encoding) into their application workflows, creating direct value for users rather than just serving internal teams.

This version also failed to achieve "product-market fit." Matt admitted that part of the reason was that the product experience was still lacking, and developers had to write cumbersome code. The real turning point came with the substantial change in the third version: Trigger.dev began executing code for users. Previously, while many users mistakenly believed that they offered execution services, developers still had to manage the infrastructure on their own. In the third version, Trigger.dev completely took over the execution aspect, providing an integrated solution from SDK to execution platform.

This transformation coincided with the rise of AI. Asynchronous, long-running tasks that may require pausing to wait for feedback ("human in the loop" or "other agents in the loop") are typical characteristics of AI Agent workflows. Trigger.dev provides a natural support platform for these types of workflows. This adjustment had an immediate effect, with the company's revenue beginning to grow robustly at a rate exceeding 30% monthly, truly finding the product-market fit.

Today, over 90% of Trigger.dev's use cases are related to running AI Agent workflows. For example, the client "icon.com" uses its platform to process user-uploaded assets, generate numerous video advertisements based on descriptions, and incorporate feedback loops in the workflow. Another EdTech client, "Magic School," uses Trigger.dev to run agents that help teachers and students plan lessons and grade assignments. These cases demonstrate the platform's core position in today's AI application ecosystem.

Open Source Strategy, Developer Experience, and the "New Users" in the AI Era

Trigger.dev is an open-source company (using the Apache 2 license). Matt explained their business model: the core open-source project contains most of the features, while commercial "cloud services" are responsible for managing complex underlying infrastructure (such as Kubernetes clusters), providing a reliable and scalable execution environment. Because most developers are unwilling to get bogged down in infrastructure management, this "open-source core + managed service" model works well.

Open source also brings unique advantages to the company. In terms of customer support, users can directly consult the open-source code repository to solve problems. Furthermore, when users discover bugs through AI coding tools like Claude and submit reports, the team can sometimes have Claude analyze the code and generate fixes directly, significantly enhancing support efficiency.

Matt specifically mentioned an interesting point: in the AI era, product users can be divided into two categories—human users and AI users. How to make AI (such as LLM) also "like" and use Trigger.dev has become a question they consider. Open source once again becomes an advantage here, as open-source projects have a broader footprint and visibility on the internet, making it easier for AI to capture them during searches and recommendations. They have even started to conceive whether it could be possible in the future to enable AI agents to autonomously complete the entire process from discovering the product, registering an account, to starting to use it.

Regarding developer experience, the team consistently adheres to the principles of "code first" and "minimal failure." They spend a lot of time discussing how to design the SDK interface, ensuring that developers can hardly make mistakes when using it. This obsession with detail forms an important part of the product's early appeal.

AI Reshaping the Team: Recruitment, Development, and Skills Assessment

The advances in AI have profoundly impacted Trigger.dev's internal operations. Matt revealed that after completing A-round financing in November 2024, the company's recruitment plans underwent substantial changes. Due to the tremendous advancements in AI coding tools (such as Opus 4.5) and cloud development environments, engineers' individual productivity has increased by "5 to even 10 times." As a result, they no longer plan to recruit engineers as aggressively as initially intended.

For the existing team, the use of AI tools has also evolved. Matt pointed out that there was an obvious division between "vibe coders" (developers with non-traditional backgrounds relying on AI for coding) and traditional developers in the early days, but this boundary has since blurred. As AI tools have become more powerful, Trigger.dev has also done a lot of work to make it friendly towards AI (such as improving documentation and building MCP servers), enabling various developers to use the product more effectively.

More importantly, AI has changed the development process and skills assessment. Trigger.dev no longer focuses on whether engineers can "manually code from scratch," but rather on how they efficiently utilize AI tools to accomplish tasks. This has become a core skill. Their hiring process includes a "trial day," where candidates are encouraged to use their preferred AI tools for work. If candidates cannot proficiently utilize these tools, they are unlikely to be hired.

In terms of code quality control, the team has also formed new methods. They acknowledge that AI coding tools can produce a lot of "rough" code. Therefore, the focus of quality control has shifted to the review stage. They utilize excellent code review tools and combine them with AI to assist in the review process. Furthermore, having a solid internal design system and a high-quality component library is crucial, as AI does not need to create components from scratch each time. Additionally, writing extensive tests has become key to ensuring the reliability of backend functions, especially when agent tasks need to verify success or failure.

Matt emphasized that the value of AI tools lies not only in "writing code" but also in acting as a "high-level engineering team," capable of quickly completing tasks such as performance benchmarking and production environment analysis, which previously consumed much time, allowing for earlier and more data-driven decision-making.

Advice for Entrepreneurs: Launch Early, Stay Close to Customers, Trust Your Intuition

Reflecting on their three-year journey, Matt and Eric offered a few core pieces of advice for new entrepreneurs about to join YC.

• Launch the product early: This is the most important advice. Only by delivering the product can one start learning what users truly want. Many insights (such as whether the product is important or if users care about service interruptions) can only be gained through launching.
• Stay extremely close to customers: Communicating with customers daily and gathering feedback can reveal many issues that are not apparent when working in isolation.
• Know when to persist: Finding product-market fit can be a lengthy process. Although Trigger.dev’s first two versions did not succeed, the team believed that the issues were worth solving, as they had personally suffered from them, so they persisted. Judging whether to continue when seeing some positive signals in the early stages but not fully succeeding is very difficult; it requires deep intuition and faith in addressing the problems.

The journey of Trigger.dev showcases a classic path of a tech startup: starting from solving a known pain point, going through multiple iterations and pivots, and ultimately, as a new technological paradigm (AI Agent) arises, standing at the core of the wave thanks to their previously accumulated product foundation. Their open-source strategy, relentless pursuit of developer experience, and redefinition of team skills in the AI era provide valuable practical insights for entrepreneurs in the developer tools field.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。