Patients with a cancer diagnosis need access to the best possible treatment as quickly as possible – but clinicians who must wade through patient records that may extend to hundreds of pages often lack the capacity to help. New York-based scale-up Triomics , which has just unveiled a £22 million Series B funding round, thinks its artificial intelligence (AI) platform can make all the difference.

I first interviewed Triomics in 2024 , when it was primarily focused on helping oncologists identify new trials and treatments that might work for patients. Since then, the business has grown rapidly – it now works with cancer centers responsible for treating around 15% of all American cancer patients – and expanded its use cases.

Still, the business’s focus has remained consistent: it aims to reduce the information processing workload that slows doctors down. “A patient’s record might be 1,000 pages long, spanning everything from handwritten notes to MRI images,” explains Sarim Khan, the CEO of Triomics, who co-founded the business with CTO Hrituraj Singh in 2021. “Before the physician even meets the patient, he or she needs to work through every detail of that record.”

Triomics’ AI platform, based on a large language model trained to specialise in oncology, can substantially relive this burden. It processes each patient’s record – including both structured and unstructured data – to produce a set of outputs that doctors can use to identify the most appropriate next treatment step. The company’s data , published with the American Society of Clinical Oncology, suggests it is possible to reduce chart review times by 67%. It has also published statistics showing a 40% improvement in matching patients to trials.

“Ultimately, this leads to better patient outcomes,” explains Khan. “New patients get the right treatment more quickly and doctors can also accelerate support for returning patients.”

One priority has been to focus on accuracy and transparency, with the platform providing sources for all its outputs and verifiable accounts of its processes, rather than requiring doctors to trust the AI implicitly. Triomics is also helping doctors to meet their compliance responsibilities, with tools that automate reporting and data submissions.

Market reach continues to grow. Existing customers include leading cancer centers such as Memorial Sloan Kettering, MD Anderson, Yale Cancer Center and the Mount Sinai Tisch Cancer Center, as well as community oncology practices such as Texas Oncology.

Singh believes the rapid growth of the company reflects the challenges facing doctors. “Oncology faces an information burden at a scale that legacy systems were never designed to handle, and that burden can stand in the way of better outcomes,” he says. ”Clinicians, research coordinators and medical assistants are working against records that have become too large and too dynamic to process manually.”

It’s an argument that resonates with Lee Schwamm, chief digital health officer at Yale New Haven Health System, an early adopter of Triomics’ technology. “This activity is labor intensive, subjective and challenging to complete in a timely manner,” he warns.

The company’s revenues are now comfortably into seven figures, but Khan and Singh are keen to go faster. Trionics now plans to expand its audit and reporting services and to make further investments in training its large language models. The business is actively recruiting in both customer-facing roles and in engineering and clinical expertise.

The Series B round will support these efforts. The $22 million raise is led by Battery Ventures, with participation from existing investors Nexus Venture Partners, Lightspeed and Y Combinator, as well as strategic backers Oncology Ventures and Precision Health Informatics, a subsidiary of Texas Oncology. Triomics has now raised total funding of more than $36 million since its launch.

Brandon Gleklen, principal at Battery Ventures, who is joining the Triomics board, is excited by the opportunity to support physicians. “Triomics has built what oncology has always needed: AI infrastructure that actually works on the full patient record,” he says.

“We are live at some of the top cancer centers and demonstrating measurable outcomes – faster enrolment, less manual chart review – and the same underlying AI infrastructure already powers multiple distinct workflows with no redundant integrations.”