Nome Uses AI to Find Rare-Disease Treatment Paths Before the Hard Work Begins

A genetic diagnosis can explain an ultra-rare disease without offering a practical next step. Nome is building an AI-assisted service to identify potential treatment routes and organize development plans for families, but identifying a mutation match is not the same as producing a therapy. The company’s test is whether it can make the work between a first screen and patient dosing less burdensome.

Nome is positioning itself as a contract research organization for smaller rare-disease groups that traditional pharmaceutical companies often do not serve. Families can upload genetic-test results; the company’s system searches for potential treatment options and returns a free detailed report after a PhD reviews the output. Founder and CEO Stevie Ringel says the AI portion of that analysis takes about 10 minutes, rather than potentially dozens of hours of manual work.

A fast screen, not a finished therapy

Nome says it has reviewed roughly 5,000 cases and now produces 80 to 100 reports a month. In about 25% of reviewed cases, it says it identified either a programmable medicine or an existing custom medical therapy that fit the known mutation. That is a company-reported indication that a route may exist; it does not establish a clinical outcome or a treatment delivered to a patient.

• A family uploads genetic-test results after a genetic disorder is identified.
• Nome evaluates possible treatment options and has a PhD review the report before returning it to the family.
• If the family hires Nome for follow-on work, the company can design clinical trials or manage a program toward drug production and delivery.

Slightly more than 10Genetic-medicine programs under Nome’s oversight

Nome says it oversees slightly more than 10 genetic-medicine programs.

The business is development coordination

Nome earns revenue when customers retain it for that later-stage work. Ringel calls the service “drug development operations”: coordinating trial design, researchers, vendors and other steps needed to move an identified option toward a drug that can be built and delivered. The company is selling an operating layer around specialized science, rather than claiming software can replace it.

The DAND Alliance offers an example of the intended role. The patient group had raised money for treatment development, but needed help deciding what to fund and in what sequence. Nome prepared a 53-page roadmap covering gaps and priorities, including animal studies, trial design, and potential researchers or vendors, according to alliance founder Jacalyn Lee.

Automation is the bet; cost is the constraint

Nome says AI currently automates about 25% of its work. Ringel forecasts that more capable AI agents could raise that share to 60% to 80% within one to two years. For drug-development expertise, Nome works with more than 80 partners, including La Jolla Labs and Dyno Therapeutics. That structure leaves the specialized development work with partner organizations while Nome handles analysis and program operations.

The financial premise is substantial, but still a goal. Ringel says customized antisense oligonucleotide treatments cost $1.2 million to $1.4 million and that Nome hopes to reduce those costs by 50%. He acknowledges skepticism that the company could add an expense to an already costly process, arguing that a dedicated coordinator can save patient groups time and money. Whether that projected saving materializes will determine whether fast screening becomes a more accessible route to personalized medicine.