From One Engineer to 6,000 Agents: How Qontext Scaled Security
Highlights
A software engineer at tech startup Qontext had built a few AI agents in-house to check their code base for vulnerabilities. That was before they signed up with RunSybil to evaluate their attack surface and experienced more than 6,100 coordinated agents engaging in the review.
Rather than spending time guiding their agents, context switching, studying their findings and cutting through the noise, Qontext’s engineer was able to sit back and let Sybil do all the work. He also was able to remediate the first vulnerability within an hour of Sybil discovering and reporting it.
That was particularly important to a cybersecurity team of one, who also is responsible for the company’s internal tooling, as well as cloud, systems, and infrastructure management.
“It was critical to have Sybil and all those agents running at the same time,” says Lars Lange, founding engineer at Qontext. “Our agents just weren’t as thorough in comparison. And working in a startup, I just didn’t have time for them. RunSybil gives us a much bigger advantage than fiddling around on our own to only come up with a worse solution.”
Unlocking Customer Growth and Meeting Audit Deadlines
Qontext, based in Berlin, Germany with 11 employees, is an AI infrastructure company, providing an independent context layer that acts as a single, central memory for enterprise AI systems.
It can be unwieldy and time-consuming for companies that use multiple AI platforms to repeatedly feed business rules, customer data, and operational knowledge into them. Qontext, founded in late 2025, unifies scattered internal data sources into a secure repository, enabling every AI agent, chatbot, and automated workflow across an organization to continuously draw from a single up-to-date context base.
To win new customers, which is critical to the success of any startup, Qontext needed to show potential customers that they had validated the security of its attack surface. And Lange was on a tight deadline since one prospective client had requested a security review before signing up to do business with them. At the same time, they were facing a critical SOC 2 audit, which is an assessment of how well they are set up to protect customer data.
Missing either deadline was not an option. And they didn’t.
Immediately after completing the first of two scheduled surface evaluations, Lange was able to give that potential customer the security validation they requested, and quickly file the needed documentation for the external audit.
“Having a breach would put a strain on acquiring new customers. That’s why we need RunSybil,” says Lange. “To have a person manually do an evaluation and go through your code, it’s just too much. It’s not going to be thorough or fast. With Sybil it’s both.”
Evaluating Qontext’s Attack Surface at Scale
Qontext, which uses both static code analysis and dynamic code analysis, started out by evaluating whether they wanted to go with a traditional, or manual, security audit, or with an AI-native approach. Lange said they decided against a traditional test because it would produce a PDF-only output with no remediation guidance. It also would have taken at least a month to run.
“You have to have trust in the provider to do a good job and we have that trust in Sybil. Without using RunSybil, I’m not sure I’d have as much confidence in evaluation findings.”
Lars Lange, founding engineer, Qontext
Once he decided to go with AI, he chose RunSybil because he valued its founders’ industry credibility, the agents’ speed and depth, and the natural alignment with Qontext’s own AI-native ethos.
“We wanted an AI-native security provider that could evaluate our platform deeply, work with code-level context, and help us move from findings to fixes quickly,” Lange adds. “RunSybil was the best fit for our priorities. As a single person working on cybersecurity, you can’t keep up with everything. But even when we’re a bigger team, I see RunSybil helping us keep that surface secure.”
He also opted to go with a white-box autonomous penetration test, enabling RunSybil’s agents to evaluate its code base, as well as its attack surface, for deeper analysis. The first test ran in September, with a second test scheduled for later in 2026.
Because RunSybil provides a real-time interface and dashboard, Lange and Qontext’s Co-Founder and Chief Product and Technology Officer (CPTO) Nikita Kowalski were able to watch as more than 6,160 Sybil agents worked on the company’s evaluation. Scale was essential because agents needed to evaluate both the surface and the underlying code base, and then triage initial findings down to actionable issues. Having live visibility enabled them to verify the depth of the coverage across the backend, frontend, and API surfaces. It also gave them extra confidence in the results.
“It was interesting to follow along,” Lange says, adding that building his own AI agents gave him insight into Sybil’s complexity. “We could verify where the agents were going and how deep they were going. We could see it wasn’t a shallow evaluation. It was good for our exec to see what the AI was doing. It’s a lot more transparent than your usual test.”
“Having a security incident is just not an option for us. That’s why we need RunSybil.”
Lorenz Hieber, co-founder and CEO, Qontext
Faster Findings and Remediation, Zero Friction
Sybil identified two high-risk and four medium-risk findings. While the high-risk issues were difficult to exploit, the impact would be severe. The medium-risk vulnerabilities, however, posed their own problems since they were easier to exploit and would have impacted customers. That means all six required immediate remediation.
Lange started making fixes in less than an hour of getting the first evaluation results, something he had expected to take days. Thanks to Sybil providing actionable remediation plans and hotfixes along with its evaluation findings, a single engineer successfully resolved all six vulnerabilities within one week.
Manually, Lange says, that likely would have taken about a month, if not longer. Weeding through the findings and creating fixes would have taken several weeks on its own.
“We’re more secure now. Our platform. Our customers. Everything is definitely more secure,” Lange adds. “RunSybil removed the manual steps of interpreting a PDF, creating tickets, prioritizing findings, and arranging separate retesting. It saved so much time.”
It also saved time when it came to submitting their SOC 2 audit report. They had the report ready within a week. Lange estimates it would have taken them one to two weeks longer without RunSybil.
The engineer also says he’s hoping to use RunSybil for long-term continuous testing. “Everything is moving fast, especially with a startup that’s growing quickly,” he adds. “There’s real value in regular testing because your surface and code base are always growing and changing. It shouldn’t just be done once a year. You need continuous testing as part of the development cycle.”
“I have an hour commute and that gives me a lot of time to think. Using RunSybil takes away the cybersecurity part of the things I worry about on the way home. Knowing it’s double checking our code while it’s checking our perimeter, that helps a lot. I can trust it’s going to be thorough and fast.”
Lars Lange, founding engineer, Qontext
Securing the Organization and Unlocking Business Growth
Securing Qontext’s attack surface and code base, simply put, means supporting the organization’s business potential.
“Companies trust Qontext with some of their most important internal knowledge,” says Lorenz Hieber, co-founder and CEO of Qontext. “We have to earn that trust, and security is fundamental to it. A breach is simply not an option. That’s why we work with RunSybil.”
The evaluations also give Lange confidence he wouldn’t have without RunSybil.
“I have an hour commute and that gives me a lot of time to think. Using RunSybil takes away the cybersecurity part of the things I worry about on the way home,” he says. “I can trust it’s going to be thorough and fast. The product makes sense and the pricing makes sense. You have to have trust in the provider to do a good job and we have that trust in Sybil.”