At a glance
WHAT IT’S REALLY ABOUT
Candid Health automates medical billing by getting claims right first time
- Nick Perry says Candid Health automates revenue cycle management (RCM) by rebuilding the medical billing stack to be automation- and AI-first for healthcare providers billing insurers and patients.
- He explains how claims are constructed (patient/provider metadata plus ICD-10 and CPT codes), how insurers adjudicate them, and why small errors can shift claims into costly outcomes like out-of-network processing.
- He argues the medical billing market is enormous (estimated roughly $200–$300B) and fragmented on the outpatient side, creating room for a new entrant that can outperform legacy, feature-heavy incumbents via automation.
- He describes Candid’s early wedge in fast-growing digital health companies, where the team could focus on submitting claims correctly the first time while integrating with other tools instead of building everything at once.
- He outlines scaling lessons: winning a competitive RFP against a long-time incumbent as a PMF milestone, maintaining culture through founder-led hiring, and expanding distribution while modularizing Candid’s automation layer to sit atop existing billing systems.
IDEAS WORTH REMEMBERING
5 ideasMedical billing complexity comes from payer-by-payer rule differences that constantly change.
He describes a core challenge as providers needing to bill hundreds to up to a thousand insurers, each with different and changing submission rules, so even the “same” visit can require a different-looking claim depending on the payer.
Denials create a costly manual workflow because payers don’t clearly explain the root cause.
He walks through how claims can be rejected (e.g., “not our patient”) or denied (e.g., no benefit), and notes providers often must do “detective work” because return codes are vague and frequently require calling insurers to understand what went wrong.
Optimizing first-pass claim accuracy is Candid’s main lever for better economics.
He says Candid’s core ethos is to “submit claims correctly the first time,” arguing it improves both net collection rate (getting more of what providers expect to collect) and cost to collect (spending less to collect).
Candid started by selling outcomes and doing manual work to learn the long tail of edge cases.
He explains the initial “product” was an API for data capture plus heavy manual operations—personally reviewing claims, submitting them, and building reporting in Google Sheets—so the team could learn edge cases and build automation informed by real workflows.
RCM isn’t a power-law problem; it’s a long-tail logic problem that demands broad coverage.
He frames RCM as “death by a thousand paper cuts,” meaning problems are distributed across many edge cases rather than a few dominant ones, which makes robust automation difficult and requires deep exposure to real claim issues.
WORDS WORTH SAVING
5 quotesWe automate RCM for healthcare providers. So we've rebuilt the entire medical billing stack that healthcare providers use to bill insurance from the ground up to be automation and AI first.
— Nick Perry
RCM is not this, like, power law where a small number of things are driving most of the pain. It is a extremely long tail of, it's a death by a thousand paper cuts problem.
— Nick Perry
I didn't realize the rest of it was they will take a brick to put the fire out.
— Nick Perry
We can't live in a world... Let me put it this way. Doctors take the oath around do no harm around patients. We should include financial stuff into that too.
— Nick Perry
At the end of the day, if you squint at each company, it's a spreadsheet. Investors invest in spreadsheets.
— Nick Perry
High quality AI-generated summary created from speaker-labeled transcript.
