The most consequential object in Bobby Reddy Jr.'s career may be a plastic hospital badge. It arrived without ceremony. Prenosis, then a young company renting a small room at Carle Foundation Hospital in Champaign, needed access. The badge happened to open nearly everything - the laboratory, the ICU, patient rooms and the institutional corridors where tidy product theories meet untidy reality. Reddy, an electrical engineer by training, treated it as an invitation. For roughly two years, he talked with doctors, nurses, technicians and patients. A company that had begun with a clever device started learning the difficult pleasure of changing its mind.
The original technology measured biomarkers from blood at the point of care. Reddy and his academic collaborator, University of Illinois professor Rashid Bashir, thought it was unusually interesting. They also thought people would want it. Several years supplied a less flattering answer. As Reddy later put it with the clean candor of someone who has paid for the lesson, nobody actually wanted what they were building.
Many founder stories edit out this bit. The unbought prototype is whisked offstage so that inevitability can make its entrance. Reddy keeps it in the story because it explains the company that followed. Prenosis did not win by insisting harder. It moved from a product in search of an application to a problem large enough to organize a decade around.
“Make sure that at the core of why you get up every single day is to solve a real problem.”Bobby Reddy Jr., AI in Chicago, 2025
The education of a badge
Reddy had always wanted to start a company. The purpose arrived later. Born elsewhere and raised in Southern California, he studied computer engineering and electrical engineering at UC Irvine, added a master's in electrical engineering, then moved to the University of Illinois Urbana-Champaign for a Ph.D. focused on biotechnology devices and algorithms. By 2012, he and Bashir were exploring the point-of-care idea. Prenosis formally began in 2014 as a university spinout, with the cheerful confidence of engineers who had made something difficult.
The hospital badge complicated that confidence. Inside acute care, Reddy encountered patients whose conditions could not be understood by one label or one measurement. The information needed to characterize their fast-changing biology often did not exist in a form a model could use. The problem was not merely to compute more elegantly over the same thin record. It was to create better data, connect it with clinical context and deliver the result while a decision could still matter.
This distinction still shapes the way Reddy talks about artificial intelligence. Prenosis does not lead with it. Hospitals have a clinical challenge; the company has built a tool intended to help evaluate it. AI combines parameters and recognizes patterns from previous patients. It is, in his formulation, a feature of the product rather than the reason to adopt it. In an industry prone to dressing ordinary software in a sequined AI waistcoat, the restraint is refreshing.
Build the missing map
Prenosis began assembling the material for a different kind of map: biological samples linked with clinical data from patients in acute care. By 2025, public profiles of the company described more than 110,000 blood samples from more than 27,000 patients. Its Immunix platform used this resource to develop diagnostic and therapeutic tools. The first commercial product, Sepsis ImmunoScore, combines as many as 22 predetermined inputs from the electronic health record, including laboratory findings, vital signs and demographic information. It produces a score and places a patient into one of four ascending risk categories.
The sober language matters. The software aids assessment; it is not meant to make a diagnosis by itself. In April 2024, the Food and Drug Administration granted it De Novo authorization, establishing a new device classification for software that assists with predicting or diagnosing sepsis. The authorization arrived after a regulatory process Reddy has described as roughly five years long. Ten days of founder folklore might fit into a launch thread. Five years of validation is more difficult to emoji.
A week after the authorization announcement, Prenosis disclosed a commercial distribution collaboration with Roche, placing Sepsis ImmunoScore on Roche's navify Algorithm Suite. TIME included the product among its Best Inventions of 2024. In 2025, Reddy was named to the TIME100 Health list. These are neat outcomes. They should not obscure the less photogenic sequence beneath them: acquiring samples, generating biomarker data, connecting hospital systems, checking performance across sites, documenting software and persuading two different audiences - clinical teams and hospital administrators - that a tool deserves space in a crowded workflow.
Reddy is particularly alert to the distance between a model that performs at one institution and a tool that travels. In conversation, he has separated generalizability from model drift: related concerns, but not interchangeable ones. An algorithm may look convincing where it was developed and falter when carried to five new sites. Prenosis's answer was to work across multiple hospital systems and to build a dataset broader than the record of any single institution. It is slow insurance against a familiar technical vanity - mistaking a local result for a universal law.
There is also the mundane matter of missing inputs. Rather than celebrate an algorithm for making do with whatever happens to be available, Reddy argues for a loop that can identify the absent parameters, prompt the appropriate tests and then evaluate the more complete picture. This view gives AI a humbler job and a more demanding environment. The model must fit into the production of reliable information, not simply consume the leftovers. Transparency matters because clinicians need to see enough of the contributing parameters to understand why a risk category appeared.
For Reddy, implementation consequently involves two negotiations. Clinicians want evidence, clarity and help with a difficult judgment. Hospital leaders must also understand workflow and economics. A technically sound model that ignores either room remains a research result. The CEO's role is to carry one coherent product through both doors.
The long middle
Reddy's passage from postdoctoral researcher to CEO was gradual. He learned fundraising, recruiting, partnerships and regulation while the technical work continued. He has said, without decorative modesty, that he is still learning the business side. The admission fits his larger pattern. He is willing to describe an unwanted product as unwanted, a slow transition as slow and AI as pattern recognition rather than sorcery.
That temperament also explains his affection for Chicago. Reddy grew up in Southern California but has spent more than two decades in the Midwest. What he likes here is the grounded quality of the people: less salesmanship, more heads-down work, fewer fireworks over the factory. It is a good city for a company whose overnight success required ten winters.
“Regardless of how tough things get, we always look to that North Star.”Bobby Reddy Jr., Illinois Grainger Engineering, 2026
In January 2026, Prenosis announced two separate $20 million commitments: a Series A led by PACE Healthcare Capital and a BARDA contract. The private round supports commercial delivery and product development. The federal contract backs work that includes an approximately 800-patient randomized trial and deployment of the Immunix platform across as many as 40 academic medical centers. The ambition is moving from diagnosis alone toward matching biological profiles with therapies and designing clinical trials around the patients most likely to benefit.
A useful kind of ambition
Reddy's stated goal reaches beyond one company. He wants to make acute care attractive to major pharmaceutical companies by showing that precision diagnostics can improve clinical trials and help connect the right therapy with the right patient. If Prenosis can demonstrate the economics as well as the clinical value, he believes the field could draw sustained drug-development attention. There is self-interest in wanting his platform near the center of that activity, and he acknowledges it. There is also a broader thesis: one company can make several products; an industry shift can make dozens.
The practical lesson from his path is smaller and more portable. Technical founders are often told to fall in love with the problem, a phrase rubbed smooth by repetition. Reddy's version restores its sharp edge. Falling in love with the problem may require admitting that the thing you already built is beside the point. It may require two years of interviews before the new direction becomes visible, a decade of data work before the product earns authorization and a willingness to sell usefulness while everyone else sells novelty.
That hospital badge opened doors, but access was not the discovery. The discovery was what Reddy did after walking through them: listen long enough for the original plan to become indefensible. Good engineering can produce an ingenious answer. Good company-building begins one uncomfortable step earlier, with the discipline to find the question worth keeping.