Skip to content
A medical claim passes through unpublished rules, a broken integration and human review before approval.

If RCM is such a killer use case for AI, why aren't there Harvey or Sierra equivalents?

Because the rules for how to submit claims correctly aren’t published, humans will still need to be in the loop.
8 min read

[Editor’s note: I invited Nick Perry, co-founder of the RCM company Candid Health, to share his perspective AI and RCM and its challenges in this article. This is one of the topics that came up while brainstorming for our upcoming Sept 22 webinar about where AI agents are working/not working in healthcare]


By Nick Perry, CEO of Candid Health

I recently read a piece on this very newsletter and had a strong reaction to it as a founder in RCM. The article was titled, “Why revenue cycle is the first ‘smash hit’ application for AI.

This isn’t the first time I’ve heard similar sentiments, but it feels like a very strong articulation of a premise I don’t agree with.

This might seem surprising coming from someone who’s been building in the revenue cycle space for years with a company that deploys AI agents. To be clear, this isn’t an anti-AI rant. AI has become a huge part of Candid (my company’s) internal operations and our products. Our second major product is a RCM agent platform.

My point of view: AI is actually really hard to apply well at scale in RCM, and it often isn’t the best tool for the job.

`

Upcoming Sessions

Live conversations with the people shaping healthcare.

View all →

Share this article

Spread the word