As artificial intelligence technologies advance, many health systems are shifting attention from specialized AI tools for individual tasks to systems that can take on end-to-end operational workflows.
Cleveland Clinic has been working with startup Luminai to automate complex administrative work, starting with areas like referral management. One of the world’s largest academic medical centers, the Cleveland Clinic reported nearly 16 million patient encounters in 2025. It serves patients at 23 hospitals and 300 outpatient facilities in locations around the world, and referrals are frequently the starting point for those care journeys.
The work of handling referrals relies heavily on manual review of faxes and manual interpretation of unstructured information.
“Healthcare’s administrative functions operate as a massive, manual coordination layer. Encoding that work into software has historically been difficult because workflows span systems and point solutions, depend on unstructured inputs, and require embedded business and clinical context at every step,” said Kesava Kirupa Dinakaran, founder and CEO of Luminai.
Recent advances in AI have made it possible to handle that complexity directly, not just automate isolated tasks, but execute full workflows reliably, he noted.
“We started with referrals because they’re one of those workflows that touch a lot of systems and involve a lot of coordination behind the scenes. There’s intake, validation, follow-up. Before this, a lot of that process depended on caregivers manually reviewing information and coordinating across different systems. In our pilots, we saw that the technology could help streamline those steps,” Rohit Chandra, Cleveland Clinic executive vice president and chief digital officer, told Fierce Healthcare.
“All of these referrals were being processed through raw fax operations. They were getting millions of faxes where the job of the operations organization was to look at every fax and the handwritten notes within them, and then read and extract both the operational and clinical data from those faxes, and then import it into the EMR [electronic medical record] to then kick off the scheduling process,” Dinakaran said. “There’s a bunch of nuance in it, all the way from just extracting basic details about the patient to identifying if the patient is a high-risk patient. Is this someone who needs urgent care? Are there schedules available for them? There is a lot of nuance and complexity to that.”
Related
Luminai built a virtual inbox agent that can triage incoming faxes to automate referral workflows. “Luminai is the first line of defense to determine if the fax is a referral and if it is, is it a high-risk, high-urgency referral? If it is, then we process those, extract the data, match it to the right provider and then kick off the scheduling process. If it’s not a high-risk patient, we still process it, get it through the queue and make sure that it’s solved for. Every fax that hits Luminai’s system gets processed in less than one minute, regardless of the volume that’s coming through, whereas in most health systems, that can take, in many cases, days, if not weeks,” Dinakaran said.
He added, “What’s interesting about that process—it’s top of funnel, and then there’s a variety of other downstream processes, things like eligibility, scheduling, post-care appointment follow-ups as well as the revenue cycle and the supply chain. There’s many of these very long-form process work that’s still happening across the admin front that Luminai has become the platform partner for these institutions.”
Cleveland Clinic’s collaboration with Luminai has been moving from pilots toward broader rollout.
“With the advances in AI, we have the opportunity to re-think and transform many of our core functions, this being one of hopefully many over time. What stood out to us about Luminai was its ability to work within complex administrative workflows in a practical way. It was about helping with real operational challenges. As we saw the results in pilot environments, we became more confident that this could scale in a way that was both useful and sustainable,” Chandra said.
While many health tech and AI companies are focused on narrow use cases like scribes and revenue cycle management, Dinakaran says Luminai has its sights set on becoming an administrative operating platform partner to large health systems.
The company launched in 2020 and signed on its first health system customer in 2023. Luminai is now working with 20 large health systems. Earlier this month, the AI-native automation platform closed a $38 million series B funding round led by Peak XV Partners (formerly Sequoia India & Southeast Asia), with participation from new investor Define Ventures and continued backing from existing investors, including General Catalyst and Y Combinator.
Luminai’s platform combines healthcare-trained AI models, a configurable workflow engine and human-in-the-loop validation. The company is using the fresh cash to expand its product capabilities, grow its engineering and deployment teams and support additional enterprise customers, executives said.
Dinakaran grew up in India and, when he was younger, was a professional Rubik’s Cube solver, holding the Guinness World Record for the most number of Rubik’s Cube solved in one hour.
“When I first came across process automation problems, it felt deeply familiar in that these are 80-step problems that should be done in five steps,” he said.
About eight years ago, Dinakaran moved to Silicon Valley and began exploring and experimenting with AI to automate complex tasks. When he was exposed to the American health system he was surprised that operations primarily still run on people, process and paper. “As I walked through the admin buildings of some of these hospitals, it just blew my mind in terms of the amount of repetitive, deeply manual, very operational work that people were doing,” he said.
“At the end of the day, there exists a set of eight to 10 core work streams that run the administration of a health system. These core work streams are actually quite interconnected, but today there’s hundreds, if not close to 1,000, point solutions solving each of these individual problems. I think the average health system in the U.S. has north of 400 point solutions for these eight to 10 core work streams,” he said. “Each of those point solutions, by nature of the amount of time it takes to set up the infrastructure, give it the context and actually kick off the automation and production, it becomes incredibly challenging to maintain.”
Related
Dinakaran maintains that the technology industry has hit the “intelligence abundance cycle” where building an AI platform that is deep yet also broad is finally possible. “What we learned was the core reason why you weren’t able to do it before was because there existed a tremendous amount of unstructured data, and so without effectively, in a scalable way, structuring this data, you can’t actually build the software layer that goes out and does this work,” he said.
Luminai brings together applied AI talent with experience from organizations such as Palantir, Cruise, Google, Coinbase and Brex, alongside healthcare operators and product leaders from Epic, Banner Health and other large-scale care delivery and health IT environments.
“Healthcare takes a long time to build, and you have to be extremely product- and engineering-obsessed and deeply embedded with your customers to be successful. We spent the last two and a half years basically doing that,” Dinakaran said.
Health systems are increasingly interested in tech platform partners that provide AI that executes multi-step workflows across systems, rather than just automating one step at a time, and also can support a range of high-impact use cases over time, Dinakaran contends.
Administrative work still accounts for up to 25% of healthcare spending. Large health systems want to automate administrative work as they face mounting cost pressure, staffing constraints and increasing operational complexity.
As the Cleveland Clinic works with Luminai, tech leadership sees opportunities to use its technology to tackle other administrative workflows across the health system, especially in workflows that are high-volume, complex and operationally intensive, Chandra said.
“While we’re being thoughtful about where we go next, the broader focus is on areas where teams are spending significant time on repetitive administrative work and where improving speed and consistency could have a meaningful impact,” he said.
Related
“There’s a transition that is starting to happen where institutions are partnering on a platform level, but it’s extremely important to be grounded in the reality of making sure that you can actually drive ROI in every use case that you choose to deploy in,” Dinakaran said. “That’s the No. 1 thing we obsess about, which is, hey, let’s solve real problems. Let’s go out and actually drive automation and task elimination in areas where real dollars are being spent today and move that to software and AI.”
He added, “The magical experiences of playing with an AI system is starting to fade and we’re coming to a place where we need to do the work of applying these in real-world use cases with the nuances of how we operate. There’s a real need for these health systems and an interest to drive that type of work that actually impacts the bottom line.”
Chandra asserts that automating complex workflows, such as referrals, doesn’t just improve operational efficiency—it also benefits providers and patients.
“Administrative workflows have a bigger impact on the patient experience than many people realize. When those processes run more efficiently and reliably, patients feel it. With referrals, faster and more consistent processing can help shorten the time between referral and scheduled care. It can also support a more responsive experience by helping ensure the right information is captured and each patient’s needs are handled appropriately as they move through the system,” he said. “Providers can feel the difference, too. When administrative processes are smoother, it can reduce back-and-forth, ease operational strain and help teams spend less time on manual coordination.”
According to Shailendra Singh, managing partner at investor Peak XV Partners, Luminai is building the intelligent orchestration layer that will define how healthcare operations function in the future.
“Their engineering rigor and customer-embedded execution model position them to become foundational infrastructure as health systems fundamentally rethink how operational work gets done,” Singh said in a statement.
