CJ Przybyl

The through-line

The consistent conviction underneath everything CJ Przybyl says is that useful technology grows out of real operational work rather than in place of it. Asked how a company can be AI-native and still employ large numbers of people, his answer is that the two are conditions of each other: "I'm a firm believer that you cannot build a lot of technology, you know, to optimize people's work unless you have people's work to optimize" [1]. That is the logic of Reserv as an AI-powered third-party administrator in property and casualty claims, and of his stated ambition to turn adjusters into sought-after AI experts rather than displace them [1].

The second thread is a bias toward doing the unglamorous version first. He describes an arc from photo-based appraisals to a claims platform to a TPA in which the operational, service-heavy layer was never treated as a compromise, and in which the insight that made the software valuable came from working inside the industry's existing plumbing [1].

On being AI-native with a workforce

The apparent contradiction between AI-native positioning and a large employee base is one Przybyl resolves by inverting it. Optimization has to have something to optimize, so the human claims work is the raw material for the technology, not an embarrassment to be engineered away [1]. His stated goal for the people doing that work is upward: he wants to develop adjusters into AI experts who are in demand [1]. The pattern is familiar from his earlier scaling experience, where taking on a contract that forced rapid growth built what he calls "this muscle memory of like scaling an organization of like, you know, service-oriented organization" [1].

On not being afraid to be a service company

He is explicit that leaning into services, rather than holding out for a pure software story, is what let the earlier business survive its long unglamorous stretch: "leaning into it and not being afraid to be a service company" produced "a decent-sized, you know, service business" that preceded the software platform [1]. He is equally clear that the two are not the same discipline. Moving from operations into a SaaS platform starting around 2015 was "just another different journey," because "building a SaaS company is totally different than building a service company," a shift he describes as fun and "a little bit terrifying as well" [1].

On finding the real problem inside incumbent systems

Przybyl's marker for knowing a business is onto something is not customer enthusiasm alone but structural understanding of the industry's software. The turning point came when the team began "trying to build software around their workflow deficiencies" and grasped how the incumbent platforms were actually assembled, including that one major vendor had built three different bespoke platforms under a single name [1]. That was the moment he felt "we're really like getting somewhere here," because "we're understanding how the fabric of the industry's working" [1]. The same instinct shows in how he got started in insurance at all: reasoning that body shop dollars flow from carriers, then cold-calling insurance executives who "just started leading us towards what they wanted, which was photo-based, you know, body, you know, appraisals" [1].

On proving demand crudely and early

His conviction that the product was real came from a deliberately unpolished test rather than a pitch deck. At a body shop conference in New Orleans, before ITC existed, the team rented a car with damage, took it to a back alley by the hotel, pulled people out of the bar, photographed the damage and showed an estimate on the phone with an app that "didn't really work" [1]. The reaction, with attendees dragging their friends over, told him "This is like real, you know, absolutely people want this" [1]. The same preference for direct contact with the work shows in how the first platform redesign was done: an internal contact center rep did the graphic design, sitting with colleagues while he whiteboarded, then testing it, a process he notes would be "rapidly compressed" with today's development tools [1]. When it improved operations and customers responded, that was the signal that the team and the thing were both good [1].

On what actually makes founders succeed

Having taught entrepreneurship at the University of Chicago, Przybyl built a course around what he thinks the textbooks miss, drawing on a class he took that was "very counter to everything I've just learned in business school" [1]. His framing to students: "it's not always about your technical foundations of business school. It's not always about being the best engineer. Sometimes it's just luck. Sometimes it's just blind ambition. It's always getting people to follow you, no matter what," alongside staying clear and a "big emphasis on pivoting" [1]. He brings in real entrepreneurs and ties the material back to academic concepts, including physics and philosophy, describing the teaching as practical and pragmatic but still rigorous [1].

He treats coming from engineering as a specific trap, because "sometimes you kind of think these other things are just easy. You can just do it," when in fact "there's a art and a science to building a business" [1]. Belief, in his account, is largely non-rational and necessary: "as any good entrepreneur, we blindly believed it was going to work no matter what," coupled with the view that he would generate his own economic returns because "who better to bet on than yourself" [1]. He is candid that this comes with cost. COVID was hard and exhausting, "a little deflating," and pushed him to look for something less volatile before he chose to jump back in [1]. He also admits misjudging the next one, telling his wife the TPA "will be an easy one. It's a TPA. It's an easy industry. It's using the software we just built," which turned out not to be the case [1].

On building a market before it has a name

Przybyl frames the early years as selling into a category that did not yet exist: "insurtech was not a thing," which meant "everyone loved to hear our story," but it also meant "a very tough ride" raising money, since "no one understood what insurtech was" and "we didn't have venture economics, so that made it hard" [1]. Deals in the segment are big and correspondingly hard to win, which he says the company survived by staying lean until the first large carriers came through [1]. Momentum, in his telling, arrived through opportunistic relationships as much as strategy, from a conversation with an insurance CEO in Milan that carried over to Zurich after the executive changed jobs, to the invitation from industry investors to start a TPA because a growing capacity provider needed one and did not want to build it [1].

Takeaways

• Reserv's premise is that AI and headcount are complements: "you cannot build a lot of technology, you know, to optimize people's work unless you have people's work to optimize" [1].
• The goal for claims staff is elevation into AI expertise, not replacement, with adjusters developed into sought-after AI experts [1].
• Services are a legitimate foundation, and he credits "not being afraid to be a service company" with building the business that preceded the software platform [1].
• Running services and running SaaS are different disciplines: "building a SaaS company is totally different than building a service company" [1].
• Conviction should come from crude live tests, like pitching a photo estimate on a damaged rental car in a hotel back alley with an app that barely worked [1].
• Real defensibility came from understanding incumbent platforms' architectural and workflow deficiencies well enough to build around them [1].
• His teaching argues success turns less on technical excellence than on luck, blind ambition, willingness to pivot, and "getting people to follow you, no matter what" [1].