When Paul Klitzke and Lukas Esch joined Strala, the company had a couple of desks, some of them assembled from monitors scavenged elsewhere in the building.
Little over a year later the company had grown close to tenfold. Its client base had multiplied by the same order, it had moved into an office built for more than a hundred desks, and it had raised a Series A in the tens of millions of dollars. Klitzke and Esch built the two functions that carried most of that weight: the claims operation and the commercial organization.
Those functions matter more at Strala than they would at most startups selling software to insurers. Founded by Timon Gregg and Armando Schmid, Strala operates as an AI-native third-party administrator, or TPA, handling property, auto and general-liability claims on insurers’ behalf. It does not sell a tool and leave the customer to deploy it. It owns the service, staffs it with licensed adjusters and answers for the outcomes. That model is the company’s central bet, and it works only with an operation good enough to honor it.
It is also what persuaded Esch to join. Many AI companies face an awkward question about what protects them as foundation models improve. Strala’s answer is that it operates inside a regulated industry and generates proprietary evidence from the claims it handles.
“We know if a claim gets reopened. We know if the claim outcome was really good and can feed this back,” Esch said.
Klitzke reached the same conclusion from a different direction. After years around venture-backed technology companies, he had avoided insurance: too complex, too slow, too few breakout companies. Strala inverted the argument for him. The regulation and entrenched process that deterred other founders were the barrier that would protect whoever cleared it, and AI had just moved the line on what could be automated behind it.
“Everything that made insurance unattractive to build in is the reason it is defensible once you do,” Klitzke said.
Building the Claims Organization Faster Than the Company Grew
When Klitzke joined, Strala’s claims department consisted of just a handful of managers and a handful of claims adjusters. He expanded it by a factor of close to ten.
The recruiting was the smallest part. New customer programs arrived with its own workflows, reporting requirements and service expectations. Klitzke built one system around the clients requirements: hiring and training pipelines, management layers between the leadership and the adjusters, and an onboarding process that could bring a large program live without improvising each time.
Because Strala runs the claims service rather than selling it, that operational work is fed directly into the product. When adjusters hit the same obstacle repeatedly, it became a requirement for engineering. When the technology improved, Klitzke redesigned how work was assigned, reviewed and closed. Few companies have the same person close enough to both halves to make that loop turn quickly.
He then joined the founders in leading Strala’s Series A, which raised tens of millions of dollars.
Turning Founder-Led Selling Into a System
Esch took partnerships and go-to-market. In the early months that meant Esch and Gregg working deals themselves. He grew the sales organization more than fivefold and moved it upmarket, toward larger and slower-moving carriers, which changed the job entirely: from winning one opportunity at a time to running a process that could identify the right buyers, diagnose their claims problems and survive a nine-month cycle.
He also built tooling that no generic sales stack provides. Alongside commercial software, Esch’s team built internal systems that scrape public sources for evidence that an insurer is struggling with claims – regulatory complaints, service failures, signs that an incumbent TPA is underperforming. Research that took an analyst days now surfaces automatically, and it points the sales team at insurers with a problem Strala is built to solve.
It is a small system with a large principle behind it: general-purpose technology becomes valuable when it is fused with specific knowledge of how an industry works. That principle runs through the product as well.
“The signal was always public. Nobody in this industry had bothered to read it at scale.”
Why Strala Hires the Veterans
The harder question was who should operate the technology. As the claims organization grew, Strala faced a choice: hire younger adjusters who would adopt new AI tools quickly, or veterans with decades of claims experience who would take longer to learn them.
Strala chose experience, and the reasoning is worth stating plainly. Claims are a business of edge cases, and no amount of technical fluency substitutes for having seen several thousand unusual files.
“You can train them on the technology, while in a week you cannot catch up on 20 years of claims edge cases,” Klitzke said.
That cuts against the assumption that AI creates value by removing people. Strala’s wager is the opposite: that software can multiply what its most experienced adjusters handle, and that this makes deep domain expertise more valuable rather than less. AI organizes the file, summarizes the documents and clears the administrative work; the adjuster spends the recovered time on the judgment calls and the phone calls, which is where claims are actually won or lost.
The result sits between two familiar models. Traditional claims services scale by adding people. Pure software companies automate what they can and hand the customer the rest. Strala is betting that regulated industries need both halves under one roof – experienced professionals accountable for the hard decisions, and technology that sharply raises what each of them can carry.
Insurance is not where anyone expected AI’s frontier. Its regulation, complexity and dependence on human judgment are exactly what has kept software companies out. They are also what makes the position defensible once someone gets in. Strala’s claim is not that AI will simplify claims work. It is that the people who understand it best have been operating without leverage, and that the company willing to hire them, own the service and build the tools around them will end up with an advantage a better model cannot erase.
Whether that holds will depend less on the models than on the operation Klitzke and Esch spent their year building.
