Did 2 Top Venture Capitalists Just Make the Strongest Case Yet for Consensus Clarity?

8 minute read

John Nebergall, Consensus’ Chief Operating Officer, Shares His Thoughts on a Recent VC Article About Faxing in Healthcare and How the Right AI-Powered Tools Can Benefit Smaller Providers

As odd as this might sound, Silicon Valley venture capital giant Andreessen Horowitz, the investment firm behind such huge tech successes as Facebook, Pinterest, Slack, Airbnb, Lyft, and Robinhood, published a forward-looking article in 2023 about… fax technology in healthcare.

The article’s co-authors, two partners at the firm who specialize in healthcare and biotech investments, acknowledge their surprise at writing about the topic. Faxing, they note, should have faded away in the 1980s, and it did for most industries.

But as they correctly point out, fax remains the dominant form of clinical document exchange in healthcare largely because these organizations are reluctant to make the massive investment required to implement new systems, add more burdens to their IT teams, and risk further frustrating their overworked clinical and administrative employees by insisting they learn new software tools.

AI Will Improve, Not Replace, Faxing in Healthcare

The VCs then argue that any new technology needs to offer a 10x improvement to displace the incumbent, especially an incumbent technology as entrenched and long-lasting as fax. And while we agree with that concept in principle, we differ with the authors’ conclusion: Because artificial intelligence represents a 10x improvement over faxing, they argue, “healthcare is going to move straight from fax machines to AI.”

Not so fast.

With all due respect to Andreessen Horowitz, we at Consensus Cloud Solutions have been in the healthcare cloud-faxing game longer than any other provider — more than 30 years, in fact — and we know from working with thousands of health organizations that faxing will remain the primary data-exchange protocol for the industry for years to come.

Yes, AI can represent a 10x improvement in faxing efficiency, speed, and accuracy, not to mention the many follow-on benefits of those improvements, including cost-savings, increased employee job satisfaction, and better health outcomes.

But just as email, Dropbox, and EHRs all failed to end faxing in healthcare, fax’s continued ubiquity in the industry strongly suggests that we should anticipate AI becoming a useful enhancement to fax technology, not its replacement.

To flesh out these ideas, we sat down with Consensus’ Chief Operating Officer John Nebergall to discuss this recent VC article and, more broadly, how AI fits into the future of fax.

Consensus Cloud Solutions:

The Andreessen Horowitz article tries to juxtapose faxing with artificial intelligence, treating them as two opposing technologies and suggesting that healthcare organizations will choose AI over fax. What do you think led them to that conclusion?

John Nebergall:

First, I should point out that we obviously disagree with the premise. The right AI tools will improve faxing processes; it’s not an either/or. 

One of the big challenges with fax transmissions — and this is probably where the authors went down the wrong path to viewing AI as fax’s replacement — is that a paper fax is unstructured data. That means a human can read and understand its contents, but a computer system can’t.

For a computer system to be able to do anything useful with, say, a patient’s lab results that come in by fax, that fax needs to be in some sort of machine-readable format so the software can identify its key details and pull them into the appropriate fields in the system. That’s where AI, and its sister technology Natural Language Processing, or NLP, can offer healthcare organizations a 10x improvement in their everyday workflows.

AI and NLP can learn to read those incoming fax documents, even if they’re written by hand, and then take the appropriate next actions, such as opening the relevant patient record and adding the fax’s key details into the right places in that record. In other words, AI software can automatically turn unstructured data (like a fax page or handwritten note) into structured data that can be ingested automatically by a provider’s digital records system.

But keep in mind, most healthcare organizations today still use faxing to exchange clinical documentation. An industry as big as healthcare, and the article’s authors themselves say it’s 20% of the US economy, isn’t going to adopt a new data exchange format overnight. That’s a big ship to steer. Overhauling a key daily workflow across this entire industry is going to take many years.

That’s why although I believe the authors are correct that AI is going to revolutionize healthcare data exchange, my prediction is that it will happen as these organizations apply AI to their existing fax workflows.

Consensus Cloud Solutions:

Let’s get specific. You’re saying AI and NLP tools can 10x the value of healthcare providers’ fax workflows. What would that order-of-magnitude improvement look like in real-world terms?

John Nebergall:

The way I see it, we can break these benefits into four main areas. 

The first is workforce optimization. Everyone in healthcare knows clinical professionals are increasingly dissatisfied with their jobs, and a big reason is that those jobs often require more time for data entry than seeing patients. One 2023 study from McKinsey & Company found that manual documentation work takes up 15% of a nurse’s shift. Not surprisingly, 45% of inpatient nurses told those McKinsey researchers that they’re planning to quit their jobs within 6 months.

Applying AI and NLP to a healthcare organization’s fax environment means those clinical professionals won’t need to spend hours every week manually reviewing every incoming fax and re-entering the data into an EHR system. Automating these common workflows will clearly improve the job satisfaction of the employees currently stuck with those repetitive tasks — and it will also enable the organization to redeploy those staff members to more productive and valuable work than data entry.

Another benefit will be speedier delivery of care. When a paper fax comes into a hospital or other care setting, nothing happens until an employee of that facility does something with it. In a busy clinic or emergency department, even an urgent fax might sit unnoticed on a fax machine for hours or days.

Applying AI to these incoming faxes can help these organizations route much-needed patient information immediately to the right people. To give you just one example, you could instruct your AI engine to scour every inbound fax for words like “critical” or “urgent” and have the AI send alerts to all relevant personnel.

A third benefit of applying AI to healthcare faxes is the ability to dramatically reduce data-entry errors. Every time an employee manually keys data from a fax into an EHR, there’s a chance for mistakes, and those chances compound as the person gets exhausted tackling a huge pile of these clinical faxes. A study in the Journal of the American Medical Informatics Association found that about 4% of lab results are entered incorrectly into EHRs. Think of all the downstream risks that flow from this inaccurate data sitting in patients’ records. It’s horrifying.

Finally, there’s the potential for huge cost savings. It costs health providers a small fortune to have people sitting at desks, manually reviewing paper documents, and then keying the contents into an EHR or another database. Turning these low-value, repetitive tasks over to an AI-powered software solution will allow these organizations to redeploy those staff members to higher-value initiatives that can generate revenue. 

Consensus Cloud Solutions:

Okay, there’s one more provocative argument in the Andreessen Horowitz article that we’d like you to address. One reason the authors cite for their optimism about AI’s potential in healthcare is that they believe the technology is on its way to essentially replacing medical professionals — or at least becoming their peers in patient care decisions. Do you agree with that?

John Nebergall:

No, I fully disagree.

The article says, for example, that “AI can already pass the medical licensing exam to become a doctor” and that “Soon, AI’s accuracy in diagnosing medical issues and recommending treatment plans will surpass humans.” The authors then conclude that “This is incredibly exciting for every person on the planet.”

I admire the idea that technologists and engineers are working to create tools that will aid trained medical professionals with the synthesis of information, but the idea that technology can replace what providers do is a pretty big leap. This is where I think we all need to step back and exercise some humility.

I don’t think we’re ready to shove trained medical professionals aside and ask ChatGPT for our child’s diagnosis and recommended treatments.

We humans have obviously been around a lot longer than artificial intelligence, and our medical professionals’ knowledge and judgment has been informed by thousands of years of learning and firsthand experience. It’s going to be difficult for a machine to beat that anytime soon. 

And that brings me back to our original question of AI displacing fax entirely. It’s important to note that not every health provider has a large EHR or other digital records system that everyone in the organization is connected to. Many smaller and specialty providers — skilled nursing facilities, post-acute-care centers, doctors’ practices — don’t have and can’t afford an EHR. But they all have some type of fax environment. That’s why it’s still the easiest way to send or receive a patient’s clinical documentation.

Do we really believe that hundreds of thousands of healthcare providers are going to give up the faxing infrastructure they’ve used every day for years or even decades for something entirely new? For at least the next several years, I think the most we can ask of AI in terms of clinical document sharing is to speed and improve our existing processes and tools — like fax — rather than hoping it will displace those trusted technologies altogether.