How Delphi Used Ranger to Scale Testing Infrastructure After Series A

The Challenge: High-Stakes Features Meet High-Velocity Development

Delphi builds digital minds for customers—AI clones that capture someone's knowledge and personality so authentically that their audience can engage as if talking to the real person. When your customer is Arnold Schwarzenegger and you're responsible for representing his thoughts and identity, bugs aren't just annoying—they're reputation-damaging.

Spencer Schoeben, Founding Engineer at Delphi, faced the classic post-Series A inflection point: increased customer expectations, expanded surface area, and zero margin for error—all while needing to maintain startup velocity.

"For the first two years of the company, we built a whole lot of different features and we were definitely in the move fast era. In June of this year, we announced our Series A and got a lot more attention. In the early days of the company, the stakes were lower and also the surface area was lower. But after having more attention on us and more paying customers, it became much more important that things didn't break and we were still able to ship quickly."

The problem was cultural as much as technical:

"We knew we needed to do more testing and internally we would always just make up new excuses for why we couldn't do it. So hiring Ranger to do it for us was pretty much essential."

The "Summer of Fire": When Moving Fast Breaks Everything

‍After announcing their Series A, Delphi experienced what Spencer calls "the summer of fire"—a critical period that forced them to fundamentally rethink how they shipped software.

"We rushed to ship a new set of features after our Series A launch. There were a bunch of bugs in the long tail of things, and it was a growing up moment. I think in the past we would ship 10 times a day and there would be bugs and then we would fix the bug and then there'd be another bug."

The consequences were real and costly:

"We have had specific examples where small bugs will lose us customers, contracts that are like $30,000 or more. So by having less bugs, we don't lose those critical customers."

The trust factor was even more critical than the money:

"We're in a place where we have high-ticket contracts but also are trying to build a consumer product where people trust us. People trust us with their identity, like Arnold Schwarzenegger. It's very important that nothing breaks. In both places, we need to not have bugs."

Why DIY Testing Failed: The Flaky Test Problem

Before Ranger, Delphi attempted to solve testing in-house. It didn't work.

"Right before we found Ranger, we were experimenting with writing Playwright tests ourselves, but we would just not do it because it was tedious and we were not good at it. People always made excuses for other things they should be doing with their time. But also even the ones we did create, they would just fail and it would often be like false positives. And so it became pretty useless."

The root causes were predictable:

"Tests were flaky because they were likely a mixture of not being written properly from the beginning and then also not being maintained as the product evolved."

Lesson learned: When your internal tests are flaky then you've lost the battle. Engineering teams need tests they can trust—or they'll ignore them entirely.

The Solution: Outsourcing Structure and Execution

What attracted Delphi to Ranger wasn't just the promise of automated testing—it was outsourcing the entire problem, including the discipline to actually do it.

"Having peace of mind and knowing that we have a QA process is really beneficial. And of course outsourcing the structure and knowledge of how to write a good test. It's really cool when things break and it's because we evolved the product and our tests evolved too."

1. Ranger Caught Critical Bugs Before Customers Did

The real validation came when tests started failing—in the right way, at the right time.

"We shipped a new feature in one area of the product and somewhere, something else totally unrelated broke and we never would have known, so we would have shipped it and customers would have discovered before us, but instead Ranger found it. Ever since that happened, it gave us the peace of mind that it does work. In our first month, we didn't have any failures. So it was sort of hard to feel like there was an ROI, but once tests actually did fail, it was pretty cool to see it caught before we actually shipped it."

One particularly critical save involved their paywall logic:

"We have customers who set different paywalls and quotas for their end users and that logic is really complicated. Ranger is able to figure out because we've created tests that specifically test these edge cases that even our own engineers don't fully understand. We were about to ship a bug that would break a paywall and make it so that pretty much nobody could talk to a Delphi. So that was a pretty critical one."

2. Engineers Focus on New Features, Not Testing Old Ones

The long-tail problem disappeared:

"We are able to focus on testing the new features without worrying about testing the entire product. So we can be hyper-focused on if we're about to launch a specific feature, spend 100% of our energy making sure that feature is delightful, knowing that Ranger will catch if something broke unrelated."

Legacy features that no one wanted to touch anymore? Covered.

3. Ranger Changed our Entire Development Culture

Perhaps the most profound impact was cultural, not technical:

"By introducing Ranger, it feels, I think even beyond the tests themselves, it has put us into a much better mindset about shipping product—that the end-to-end tests are a critical part of the flow of building a new feature."

Before Ranger, Delphi didn't even have a staging environment:

"Before Ranger, we didn't really have staging at all. Starting to work with Ranger was part of an effort to just create a better development cycle in general. I think Ranger was a critical piece of that, but it wasn't the only piece. It helped us align with best practices around shipping to staging first. We now try to much more carefully only ship to actual customers in the middle of the day after all tests have passed and been triaged. I think that alone has also been really valuable, just making sure that we're not so chaotic."

The result? A complete transformation in how they ship.

The Bottom Line

For high-growth companies managing complex products with high-profile customers, the "move fast and break things" era has an expiration date. The question is: how do you maintain velocity while eliminating the chaos?

The numbers tell the story:

• $30,000+ in contracts saved by catching bugs before customers
• 2x team growth without proportional increase in bug frequency
• 10x daily deploys reduced to 1x with higher confidence and faster net progress
• Zero engineering time spent writing or maintaining end-to-end tests
• Complete peace of mind that legacy features stay stable

Delphi didn't just add testing—they fundamentally transformed how they ship software. The result is a development culture where speed and stability aren't trade-offs, but complementary forces.

Ready to graduate from "move fast and break things"? See how Ranger works for your team:

Book a demo →