Another AI startup helping suppliers automate administrative processes secured a major fundraising round on Thursday.
Based in San Francisco Lumine hung $38 million in Series B funding, bringing its fundraising total to $60 million since its inception in 2020. The round was led by Peak XV Partnerswith the participation of General catalyst, Y combiner And Define businesses.
The company seeks to replace fragmented and manual healthcare administrative workflows with an easy-to-manage AI system.
Luminai CEO Kesava Kirupa Dinakaran said the startup’s platform is built using 3 layers. The first is data transformation.
The AI engine first converts messy healthcare documents such as faxes and PDFs into structured, usable data by distinguishing relevant clinical information from irrelevant noise, Dinakaran explained.
“For example, when it comes to referrals, when a health system receives faxed documents, only a fraction of them are true referrals. The rest could be pizza flyers, sales spam, provider thank you notes, and dozens of other types of documents mixed in. Traditional software cannot determine what a pizza flyer is versus a referral. Luminai’s model is able to identify what a referral is, extract the relevant clinical and administrative information and transmit clean, structured data that can be read by software,” he noted.
The second layer is a knowledge graph that learns how each health system works, including its routing rules, policies, exceptions and institutional judgment, Dinakaran said.
This means that when a referral arrives, the system can automatically determine where it should go – even in complex cases where routing decisions are not explicitly documented – by replicating the judgment of long-time staff who would otherwise rely on their experience and institutional memory, he noted.
“Each health system has its own processes and policies,” Dinakaran said. “Luminai encodes this operational logic into versioned, verifiable infrastructure, rather than relying on ‘tribal knowledge’ that lives in people’s heads. »
He explained that the third layer, workflow execution, uses agents to then classify documents, associate them with the right patient and provider, route them to the right department and trigger any necessary downstream actions.
When the system is not safe enough to act autonomously, it routes work to a human operator with all the context already in place, Dinakaran added.
But this only happens in rare cases. He said one of the key benefits the startup’s platform offers customers is reducing the administrative burden on healthcare teams.
“There could be more than 100 people whose entire job is processing incoming faxes in a health system, work that no one signed up for, but that must be done for the institution to function,” Dinakaran noted.
The company is currently in partnership with Cleveland Clinicwhich uses the platform to automate complex administrative workflows, starting with reference management.
Referrals are just the first in a long list of use cases, according to Dinakaran. He said the partnership is designed to expand into “high-volume, operationally-intensive workflows” in the administrative domain.
As the collaboration moves from pilot to broader deployment, it reflects a larger shift: health systems are seeking AI platforms that can support end-to-end operational workflows, not just help complete individual tasks.
Photo: sesame, Getty Images
