No More Flaky Tests: Propolis Is Revolutionizing Quality Assurance

Propolis is an AI-powered quality assurance (QA) platform that automates testing by simulating real users through swarms of autonomous browser agents. Founded in 2025 and based in New York, the startup emerged from the Spring 2025 Y Combinator batch with a radical proposition: hands-off QA. By replacing manual testing and brittle test scripts with intelligent agents that behave like real users, Propolis offers engineering teams a way to test their applications faster, more thoroughly, and without sacrificing valuable development time.

Traditional QA processes, whether manual or automated, are struggling to keep pace with the accelerated speed of modern software development. Propolis addresses this gap by taking an entirely new approach—deploying synthetic users who can independently navigate, test, and even communicate insights about software applications in real time.

How Did the Founders' Backgrounds Shape the Vision for Propolis?

Marc Papazian and Matt White, the co-founders of Propolis, bring an impressive blend of real-world engineering experience and unorthodox life pursuits to the startup. Marc, a former product lead at Palantir and director at an AI startup, has a résumé that includes coordinating mission-critical deployments like the federal pandemic response. Outside of tech, he’s a winemaker and a beekeeper—both fields that require patience, precision, and systems thinking.

Matt White, on the other hand, brings deep infrastructure knowledge from his time as a senior engineer at Airtable. There, he focused on the types of tools and systems that reduce the need for traditional QA. Together, Marc and Matt have seen firsthand how much time developers waste writing, maintaining, and debugging tests—time that could be spent building better products.

Their shared frustrations with the status quo and their mutual interest in systems that learn and adapt were the perfect ingredients for launching Propolis.

What Problem Does Propolis Solve in Modern Software Development?

Modern engineering teams face a trifecta of QA-related challenges:

1. Wasted Engineering Time
   Developers often have to write and maintain complex test suites, slowing down the development process and draining resources.
2. Delayed Releases
   Manual QA and flaky automated tests contribute to release bottlenecks, forcing teams to wait on test results or fix broken scripts before shipping.
3. Bugs Reaching End Users
   Even with rigorous QA processes, bugs often slip through the cracks and impact the user experience, leading to support costs and reputation damage.

These issues scale exponentially as companies grow. Some enterprises throw entire teams at the problem, while others accept the harsh reality of “the users will find the bugs.” Propolis challenges that assumption head-on.

How Does Propolis Work?

At the heart of Propolis is a swarm of AI-driven, autonomous browser agents. These agents mimic real user behavior, learning the interface, navigating workflows, and stress-testing applications automatically. Here’s how it works:

• Easy Setup: Teams provide a URL and, if necessary, credentials. No test writing, no manual setup.
• Exploratory Coverage: The agents explore every corner of the application, discovering user paths that even the product team might miss.
• Autonomous Communication: The swarm communicates internally, sharing findings and dividing testing responsibilities.
• Bug Reporting with Context: When an issue is detected, Propolis doesn’t just flag it. It delivers a complete bug report with:
  • Video recordings
  • System traces
  • Reproduction steps
  • Integrations into platforms like Slack, GitHub, and Linear

The result? Real-time insights that eliminate the guesswork and human error associated with manual QA.

Why Is the Browser Agent Swarm Approach So Revolutionary?

Most automated testing systems today rely on pre-scripted actions—click here, expect this, validate that. The problem? Those scripts break as soon as the UI changes or a new feature is added. Teams spend as much time fixing tests as they do writing them.

Propolis flips this model on its head. Its agents don’t need scripts because they adapt. Like human testers, they learn the application by interacting with it, but unlike humans, they scale instantly and never get tired. Whether it's a checkout flow, a settings menu, or a hidden feature toggle, these agents find and test it.

By building a distributed intelligence system for QA, Propolis has created a testing method that’s not only more robust but also self-healing and scalable.

How Does Propolis Integrate with Engineering Workflows?

Propolis was designed to plug directly into modern developer tools and workflows. Whether a team is using GitHub for version control, Linear for issue tracking, or Slack for team communication, Propolis delivers bug reports where they’re already working.

Its deep integrations and automated documentation of findings allow teams to go from bug discovery to bug fix with minimal friction. Developers can even trigger a swarm test as part of a CI/CD pipeline, turning QA into a continuous, invisible layer of protection rather than a stopgap at the end of the dev cycle.

Who Is Already Using Propolis and Why?

Propolis has already found traction among fast-paced engineering teams who value speed and stability. Startups pushing weekly or even daily code releases rely on Propolis to maintain quality without sacrificing velocity. For these teams, QA isn’t just a checkbox—it’s a strategic advantage.

By replacing tedious test writing and manual regression testing with intelligent automation, Propolis helps developers reclaim time and confidence. As Marc and Matt like to say, Propolis helps you “test like it’s production—because it is.”

What’s Next for Propolis?

While Propolis currently focuses on web applications, its core architecture is designed to be extensible. Future plans may include:

• Mobile App Support: Expanding browser agent capabilities to test mobile-native applications.
• Deeper Semantic Understanding: Using large language models to interpret product documentation, changelogs, and design specs to inform testing behavior.
• Real-User Feedback Integration: Combining synthetic testing with anonymized real-user telemetry to prioritize coverage and bug fixes.

Propolis’s mission is to eliminate QA bottlenecks entirely, and they’re just getting started. With a small, nimble team and a sharp product-market fit, the startup is positioned to redefine how software teams think about testing.

Why Is Propolis More Than Just a QA Tool?

Propolis isn’t just a testing platform—it’s a philosophy shift. It challenges the notion that QA must be slow, painful, and expensive. By introducing autonomous intelligence into the development cycle, it offers a glimpse into the future of human-AI collaboration in software engineering.

In a world where software changes every day, Propolis ensures QA can keep up, not by hiring more people, but by deploying smarter systems. It’s fast, adaptive, and surprisingly elegant. For teams who want to ship faster without compromising quality, Propolis is not just a tool—it’s a secret weapon.