Probably Genetic recently raised $12 million to expand its rare-disease platform and move further into drug discovery. What interests me about this development is not the funding itself, but the business model behind it. The company is trying to connect genetic diagnosis with pharmaceutical research, and I think that connection could become more important than the diagnostic technology alone.

Probably Genetic uses digital assessments and genetic testing to help identify people who may have an undiagnosed rare disease. Its latest funding, reported by Fierce Healthcare, will support the expansion of this platform and its move into drug discovery.

I see a real problem worth solving here. Genomic medicine has become increasingly good at reading DNA, but the path from sequencing a patient to giving them a useful answer is still surprisingly inefficient. Patients can spend years moving between specialists, accumulating diagnoses that describe their symptoms without explaining their cause. Even when sequencing is available, finding a relevant variant does not necessarily mean that the variant can be interpreted confidently or that the result will change treatment.

To me, this is one of the more important bottlenecks in precision medicine. We have invested heavily in the ability to generate biological data, but much less attention is often given to the infrastructure needed to connect those data to patients, clinical decisions and treatments.

Probably Genetic is interesting because it approaches diagnosis as a problem of patient identification as well as laboratory analysis. Digital assessments can help identify people who may warrant genetic testing, while partnerships with laboratories and pharmaceutical companies can create routes to testing and research that might not otherwise exist. The company is trying to make the journey from a suspected disease to a molecular diagnosis easier to navigate.

I think that is a sensible direction, although I would be careful not to confuse easier access to testing with better diagnosis. A questionnaire cannot establish a genetic disease, and sequencing does not always produce a definitive answer. Variants of uncertain significance remain a problem, some clinically relevant changes are difficult to detect with standard testing, and a negative result does not necessarily exclude a genetic cause. The quality of the clinical information surrounding a sequence can be just as important as the sequence itself.

What I find more ambitious is the company's interest in drug discovery. A diagnostic platform can become more valuable if it helps identify not only who has a disease, but also how that disease presents across different patients. Linking genotype to phenotype could help researchers define patient subgroups, recruit participants for clinical trials and investigate why people with apparently similar diagnoses progress differently.

This is where I think the opportunity extends beyond rare-disease testing. A well-structured dataset linking genetic variants to detailed clinical phenotypes and longitudinal outcomes could become a useful research resource. It might help pharmaceutical companies find patients for diseases that are difficult to study because the affected population is small and geographically dispersed.

But there is an important distinction between a large dataset and a scientifically useful one. Genetic associations do not automatically explain disease mechanisms, and patterns in clinical records can reflect how patients were referred or tested rather than the underlying biology. A platform may identify a patient with a likely pathogenic variant without revealing which intervention would work. Moving from diagnosis to drug development still requires functional evidence, a plausible therapeutic target and clinical validation.

I would therefore judge Probably Genetic less by the number of people who complete its assessments and more by what happens afterwards. How many patients receive a confirmed molecular diagnosis? How long does the process take? Does the result alter clinical management? Do pharmaceutical partnerships lead to better trial recruitment or meaningful discoveries? These are more important questions than the number of records accumulated.

There is also a commercial tension that I find worth watching. Pharmaceutical companies have a clear incentive to identify patients who might benefit from their treatments, and that funding could make testing available to people who might otherwise struggle to access it. At the same time, patients contributing genetic and clinical information should understand how their data may be used and who may benefit commercially. Building trust will be part of building the platform, not an administrative detail to address later.

My view is that Probably Genetic is working on a problem that deserves more attention, but its long-term value will depend on whether it can connect diagnosis to better clinical and scientific outcomes. I am less interested in another company claiming to apply AI to genomic data than in companies that make the entire process work more effectively, from identifying a patient to helping determine what should happen next.

If Probably Genetic can establish that connection, its most valuable asset may not be the testing platform itself. It may be the network linking patients, genetic evidence, clinical information and drug development. That is a much more interesting proposition than simply making genetic testing easier to access.