“Our Day Starts With Testkube”: A CTO’s Perspective on Testing at Thoras

I recently caught up with Reilly Herrewig, CTO at Thoras, to dig into how his team tackles testing for a product where failure simply is not an option. What really stood out was just how deeply testing is woven into their engineering culture. It is not just a box to tick before shipping; it is how they kick off every single day.

But before we get into why testing is so critical for Thoras, let us take a quick look at what they do.

What Thoras does

Thoras is a Kubernetes workload management platform that optimizes efficiency and reliability by forecasting usage, right-sizing compute resources, and making proactive scaling decisions. It is not just observing clusters; it is in the critical path.

"We are actively managing our customers' production workloads," says Reilly Herrewig, CTO at Thoras. "If there is a problem with that, it could impact our customers' product stability." That level of responsibility shapes the team's engineering culture, especially in how they test, validate, and harden every change before it reaches customer environments.

Why testing is non-negotiable

For most software companies, a bug in production is just an inconvenience but for Thoras, it can have a direct impact on their customers' infrastructure.

Because their platform makes active decisions about customer infrastructure, any regression in their system ripples outward. A miscalculation in resource prediction could cause applications to throttle. A bug in their scaling logic could trigger unnecessary scale ups that inflate cloud costs. In the worst case, a failure in their core engine could contribute to downtime for the applications they are supposed to protect.

"The impact to our customers if Thoras was not doing its job could be significant," Herrewig says. "Being as confident as we can possibly be that the product is rock solid and that we are being vigilant in our search for regressions is absolutely critical."

That need for confidence is what pushed Thoras to invest heavily in end-to-end testing, and what led them to Testkube.

When traditional test runners fall short

Validating a product like Thoras requires more than unit tests or simple integration checks. It means spinning up realistic cluster states, deploying actual workloads, applying complex resource configurations, and verifying multi-step system behaviors end to end. Traditional CI runners are not built for that. They execute scripts; they do not understand Kubernetes.

"We needed a tool that would allow us to orchestrate complex end to end tests for Kubernetes workloads," Herrewig explains. "Since Testkube is literally built for running on Kubernetes, it was really the perfect fit."

Why not build it themselves?

As a Kubernetes-focused company, Thoras has deep expertise in the platform. Their engineering team lives and breathes Kubernetes every day. They know exactly what a serious testing infrastructure needs to do and they could have built their own from scratch.

They gave it a lot of thought, but in the end, they chose to spend their precious engineering cycles building features for their product and customers.

"We absolutely could have built our own solution," Herrewig says. "But why do that when someone else has already done that? We can partner with them to short circuit all that work and focus more on delivering Thoras to our customers."

Every hour spent maintaining a homegrown test platform was an hour not spent improving the core product. Thoras's edge comes from their prediction engine and optimization algorithms, not from building test infrastructure. Testkube gave them a production-ready foundation, minus the maintenance headache.

"A lot of times, specialized tools that focus on one specific thing can be a lot better than a tool you built internally when you are sharding your time across maintaining it plus a hundred other things," Herrewig adds.

You are likely already orchestrating tests. Here is when it makes sense to invest in a dedicated platform, and what that actually looks like in practice. Read: You are already doing test orchestration →

A daily ritual built on Testkube

Today, Testkube is baked right into how the Thoras engineering team works.

Because their end to end tests are sophisticated and infrastructure intensive, they do not run on every commit. Instead, the team runs them on demand throughout the day and as part of a nightly suite. The results feed into a quality analysis report that gets pushed to Slack each morning.

"We start every morning by reviewing our Testkube driven quality analysis report," Herrewig says. "It is become a core part of our engineering culture."

That daily ritual gives the team a shared view of system health. When the report is green across the board after a day of shipping new features, everyone knows the product is solid and ready to go. If something is off, they catch it before it ever hits a customer environment.

"It gives us the opportunity to identify regressions early," Herrewig explains. "That obviously protects our customers, protects our product's reputation, and it gives us confidence to enable rapid delivery."

"It is just a great way to start your day, especially when we shipped a bunch of product the day before and it is all looking green and good."

For the Thoras team, testing is not an afterthought; it is part of the delivery process itself. Every new feature ships with its own end-to-end tests, and those tests are treated just as seriously as the feature code.

"Just as important as the implementation of the feature is the implementation of our testing," Herrewig says. "Testkube is a part of that workflow and that delivery model for us."

Scaling testing alongside engineering velocity

Like most engineering teams today, Thoras uses AI tools to speed up development. Code ships faster, features land sooner, but that kind of velocity brings its own challenge: testing has to keep up.

"Agents and LLMs are helping engineering teams ship more product, faster," Herrewig observes. "Your testing maturity and your testing investments have to scale to meet that velocity. You gotta scale them together."

For a company like Thoras, where shipping a bug can have real consequences, keeping that balance is crucial. Moving fast only matters if you can keep the reliability your customers count on.

Testkube gives them the infrastructure to do just that. As engineering output ramps up, their testing scales right alongside it, so speed never comes at the cost of stability.

AI coding assistants help your team ship faster, but faster toward what? How to build a testing strategy that keeps pace without breaking what you have already shipped. Read: Why AI will not replace engineers, but it will break your QA strategy →

Built by Kubernetes experts, for Kubernetes experts

One thing that really stood out to Herrewig during the evaluation was that Testkube feels like it was built by people who truly get Kubernetes.

"We are a team of Kubernetes nerds," he says, "and it is very clear the Testkube team is too. That expertise shows in the product."

For teams building Kubernetes-native software, that shared understanding makes a big difference. Generic testing tools often mean workarounds and compromises. They were not built with Kubernetes in mind, so teams end up fighting the tool instead of focusing on their tests.

Testkube speaks the same language as the systems it is built to test. For a company like Thoras, where everything is built on deep Kubernetes expertise, that alignment is a game changer.

Advice for engineering leaders

For CTOs and engineering leaders evaluating their testing strategy, Herrewig's recommendation is straightforward.

"If Kubernetes is foundational to your platform, you want tools built by people who deeply understand Kubernetes. Testkube absolutely reflects that expertise."

He continues: "I would recommend Testkube to any engineering leader looking to build out a robust testing platform without having to reinvent orchestration."

The alternative, building and maintaining a custom solution, comes with real costs. Not just in initial development, but in ongoing maintenance, debugging, and the opportunity cost of engineering time spent on plumbing instead of product.

"Testkube allowed us to bypass all the platform plumbing and focus on building Thoras."

Putting it all together

For Thoras, Testkube is not just another testing tool. It is the foundation that lets them ship a mission-critical product with confidence.

Their platform manages production workloads for customers who need both reliability and cost efficiency. That kind of responsibility demands rigorous testing. Testkube gives them the infrastructure to validate complex Kubernetes behaviors at the depth their product needs.

Their mornings start with a Testkube report. Every feature ships with Testkube tests. Their velocity scales because their testing infrastructure scales right along with it.

"We are huge fans," Herrewig says. "Testkube is a core part of our workflow and a big reason we are able to deliver quickly and confidently."

Key takeaways

• Testing is non-negotiable when your product is in the critical path. Thoras actively manages customer production workloads, so a regression in their system ripples outward into customer reliability and cloud costs. That responsibility drove the entire engineering culture around testing.
• Traditional CI runners cannot validate Kubernetes-native products. Thoras needed to spin up realistic cluster states, deploy actual workloads, and verify multi-step system behaviors. CI runners execute scripts; they do not understand Kubernetes.
• Buying beats building when the tool is not your product. Thoras had the expertise to build their own test orchestration platform. They chose to spend engineering cycles on their prediction engine instead, which is where their competitive edge lives.
• A daily testing ritual changes the engineering culture. Thoras starts every morning with a Testkube quality report in Slack. That shared view of system health protects customers, protects product reputation, and gives the team confidence to ship quickly.
• Testing has to scale with AI-assisted velocity. As AI tools accelerate code generation, testing infrastructure has to scale right alongside it. Speed only matters if reliability holds.

Frequently asked questions

Why did Thoras choose Testkube for Kubernetes testing?

Thoras chose Testkube because it was built specifically for running tests on Kubernetes. As a Kubernetes workload management platform that actively manages customer production infrastructure, Thoras needed a testing tool that could orchestrate complex end-to-end tests against realistic cluster states, deploy actual workloads, and verify multi-step system behaviors. Traditional CI runners execute scripts but do not understand Kubernetes. Testkube was a natural fit because it was built by Kubernetes experts for Kubernetes experts.