Patient referrals to medical specialists can be a complicated, highly manual and slow process, leading to delays in patient care, payer denials and missed patients.

When a primary care doctor refers a patient to a specialist, those referrals come in by fax, email or e-portal, and quickly pile up.

Some healthcare professionals refer to this as the "black hole problem," as patients can fall out of the system and there's no visibility into the patient healthcare journey.

Health tech startup Tennr developed an AI model to fix the front office patient referral system by automating workflows and documentation intake. Tennr built language models designed to read referrals, parse through documents, extract relevant information and route it appropriately and then automate workflows, reducing error rates and processing times. 

Tennr says it developed the only vision-language model trained to interpret unstructured medical records against complex payer criteria.

The New York City-based company, founded in 2021, says it now processes 10 million documents a month, and that's growing. Tennr works with hundreds of healthcare organizations that host thousands of providers, according to co-founder and CEO Trey Holterman.

The goal is to streamline the pre-visit process to boost patient conversions for providers and reduce denials while also improving the patient experience, according to executives.

The startup is now armed with more funding to expand its products and grow its business with more providers. Tennr raised $101 million in a series C round led by IVP. New and existing investors including Andreessen Horowitz, Lightspeed, GV, ICONIQ, Foundation Capital and tech executive Frank Slootman, former CEO and chairman of Snowflake, also backed the series C round.

The funding round boosts Tennr's valuation to $605 million, the company confirmed. 

Tennr says it has more than tripled its revenue since its $37 million series B round just two quarters ago, in October. The company pocketed $18 million in a series A round in March and has raised $162 million total to date.

The startup, which hasn't "touched" its series B cash yet, according to Holterman, is focused on continuing to build out a "world class" team of engineers along with its operations expertise around billing criteria and referral management, he said. Tennr also will use the funding to fuel its go-to-market strategies.

The company is going after "hairy" documentation reviews that almost every specialist provider has to tackle, Holterman said.

Holterman met co-founders Diego Baugh and Tyler Johnson at Stanford University as engineering students working on advanced AI and large language model research. Holterman said he learned about the "black hole" of the referral maze from his mother, who while working in family medicine, showed him how chaotic and slow the handoff between providers could be. 

"The moral of the story for me is, always just listen to your mom," he said during an interview with Fierce Healthcare.

Baugh experienced these challenges personally as a patient when six-week delays between GI appointments sent him to the emergency room in college.

Each year, more than one-third of Americans are referred for specialty care, imaging, equipment, or treatment. But many of those patients fall through the cracks. A 2017 study found that 50% of all specialty referrals aren’t completed due to miscommunication, misdirected referrals or missing information.

It’s also estimated that 70% of healthcare organizations still use fax machines to exchange medical information.

While many startups have been trying to digitize faxes out of existence, Tennr is focused on integrating artificial intelligence with the tools that providers already use. But the company isn’t focused on fixing the fax problem in healthcare, it’s more broadly focused on improving the pre-visit patient processing problem. 

Streamlining this process will reduce patient delays and denials across the U.S. healthcare system and help providers to care for patients before their health problems escalate to require more expensive, higher acuity hospital care, according to the company.

“Forcing healthcare providers to change the way they refer their patients doesn’t work. Many have tried. Tennr is the first company that works the way healthcare already does: no EMR rip-outs, no need to retrain providers, no changes to how documentation is shared. By combining deep customer empathy for specialist workflows with technical excellence, Tennr builds software that actually gets used because it works with the system, not against it,” said Zeya Yang, partner at IVP, in a statement.

Tennr built RaeLM, its proprietary vision-language model trained on 100 million anonymized healthcare documents, 2.3 billion distinct data fields and 8,000 sets of criteria. Unlike generic large language models, RaeLM is optimized to understand the nuanced data in clinical notes, scanned forms and checkboxes, according to the company. It evaluates documents against complex payer criteria to flag potential denials and ensure cleaner submissions from the start.

Along with the fresh funding, the company also announced the launch of Tennr Network that connects referring providers, receiving providers and patients. With the network, Tennr aims to give patients and providers more real-time visibility into the referral status.

"You can run a perfect operation, and you're still going to have some patients fall through the cracks because they lack visibility, they lack clarity on financial responsibility, and ultimately, you're waiting on a payer to review a prior auth," Holterman said. "We're launching our Tennr Network product, which basically provides complete visibility into where a patient is in the process of getting care. It's a great set of tools for our customers to be able to share with the senders of patients and the patients themselves."

"In the early tests that we're seeing, there's a profound increase in patient adherence, in patients getting from point A to point B and service satisfaction for patients," he added.

Through the Network product, referring providers can see the current status of every patient they’ve sent out, eliminating phone tag and guesswork, while receiving providers can track the status of every referral, see which need more documentation, and identify which sources are driving the most conversions.

Patients also can see when their referral was accepted, when it's scheduled, and what to expect to pay.

A year ago, Tennr was focused on the patient intake and documentation review process but has since expanded its capabilities. "We believed that if you could structure the data, if you could fix the reading of these documents problem, you'd be really well set up to go do a bunch of other work for the practices. And we've done that. We built an eligibility benefits product, we built a qualifications billing engine product, we built a communication coordinator. We're building a patient experience product and an auth review product. We've basically structured all this documentation and compounded that into a bunch of product lines," Holterman said. 

Ty Barnett, CIO at durable medical equipment supplier Norco Inc., said Tennr "revolutionized" its fax-to-intake workflow, "eliminating hundreds of hours of manual effort each day, removing human errors, and accelerating the creation of patient intakes."

"We’ve redefined operational agility in our revenue cycle—it’s not just about moving faster—it’s about serving healthcare practitioners and patients more effectively, in alignment with our mission of Serving You Better,” Barnett said in a statement.

Editor's Note: This story has been updated to reflect Tennr's valuation post its series C funding round.