CHAPTERS
- 0:05 – 0:26
Candid Health’s core product: automating revenue cycle management (RCM)
Gustaf asks what Candid Health does, and Nick explains the company’s focus on automating the medical billing workflow for healthcare providers. He frames it as rebuilding the billing stack to be automation- and AI-first.
- •Candid automates RCM for healthcare providers
- •The product is positioned as a rebuilt “medical billing stack”
- •Automation- and AI-first is presented as a foundational design choice
- 0:26 – 1:05
Why getting paid is hard: hundreds of payers and constantly changing rules
Nick describes the fundamental challenge for providers: they may need to bill hundreds to a thousand insurance companies, each with different submission rules. Even identical services can require different claim details depending on the payer, creating a moving-target compliance problem.
- •Providers must bill many insurance companies with different claim rules
- •The “same” visit can require different claim formats across payers
- •Staying current on payer rule changes is a major source of complexity
- •The difficulty is in keeping an automated system aligned with shifting requirements
- 1:05 – 2:12
What patients see vs. what happens behind the scenes: copays, deductibles, coinsurance
Gustaf asks what’s on a medical bill and who pays; Nick walks through a simple version of the payment flow. He explains what can be collected upfront versus what’s determined later after the insurer applies benefits.
- •A patient may pay a copay upfront, with additional amounts determined later
- •Claims are created after the visit and submitted to the insurer
- •Insurers apply benefits to determine deductible/coinsurance responsibilities
- •Timing can range from weeks to months before patients know final amounts owed
- 2:12 – 3:16
How a claim is constructed: claim forms, patient/provider data, and code sets
Nick explains what actually goes onto an insurance claim and how it’s submitted, noting most claims are electronic even though paper forms exist. He outlines the mix of administrative “metadata” and clinical/service codes that ultimately drive reimbursement decisions.
- •A physical claim form exists, but most claims are submitted electronically
- •Claims include patient, provider, location, and insurance details
- •Clinical/service details are represented via ICD-10 and CPT codes
- •Front-desk insurance capture feeds required claim fields before submission
- 3:16 – 5:17
How insurers adjudicate claims: rejection, network status, and benefit application
Nick breaks down what happens once an insurer receives a claim—first verifying the patient and then applying network rules and plan benefits. He emphasizes how minor data issues can have major consequences, such as claims being treated as out-of-network.
- •Insurers first determine whether the claim belongs to their member (rejection vs. continuation)
- •Network determination can be sensitive to small errors like address formatting
- •In-network vs. out-of-network adjudication changes patient financial responsibility significantly
- •Coverage varies across plans, including employer self-funded vs. fully funded structures
- 5:17 – 6:18
Why payment can take weeks: operational bottlenecks and manual billing work
Gustaf asks about timelines; Nick describes best- and worst-case payer response times and highlights provider-side delays. He points to backed-up billing departments and manual, “clicky” workflows as key reasons the process drags out.
- •Fastest payers can adjudicate in about a week and a half; slow cases can take weeks
- •Provider-side coding and submission can lag due to queues and staffing constraints
- •Even with fast follow-up, slow payer cycles extend timelines
- •Manual processes contribute heavily to delays
- 6:18 – 7:27
When claims go wrong: denials, vague codes, and “detective work” with payers
Nick explains the heavy labor involved in resolving rejections and denials. He describes how response codes from payers can be too vague to act on, forcing providers to call insurers, investigate, fix issues, and decide whether to appeal.
- •Denials/rejections trigger a time-consuming investigation process
- •Returned codes are often too high-level to identify the real issue
- •Providers frequently must call payers and wait through long hold times
- •Fixes can involve getting updated insurance from patients or filing appeals
- 7:27 – 10:03
Sizing the opportunity and choosing a starting point in a fragmented market
Nick discusses market size estimates for medical billing and distinguishes inpatient versus outpatient segments. He explains why outpatient was more fragmented and how Candid used the digital health wave as a go-to-market wedge to focus on core claim submission.
- •Medical billing market is described as roughly $200–$300B depending on estimates
- •Inpatient is dominated by a few large players; outpatient is fragmented across many systems
- •Replacing incumbents is hard due to feature breadth and entrenched workflows
- •Candid rode digital health/telemedicine growth to start with customers seeking a “Stripe for medical billing”
- 10:03 – 11:00
Candid’s core ethos: submit claims correctly the first time
Nick explains why Candid initially focused narrowly rather than building every adjacent workflow. He describes reverse engineering payer rules and embedding them into the platform so claims go out correctly on the first submission.
- •Focus was narrowed by integrating with existing tools instead of building everything (e.g., payments)
- •Candid prioritized the initial insurance submission as the core billing step
- •The approach relies on reverse engineering payer rules and maintaining them over time
- •Correct first submission reduces downstream rework and operational load
- 11:00 – 13:00
Landing the first customer: credibility, urgency, and a “hair on fire” problem
Gustaf asks why the first customer trusted them; Nick points to personal credibility and the severity of the RCM pain. He also reframes insurer-provider dynamics as differing incentives rather than constant intentional deception, while emphasizing the challenge of ever-changing rules.
- •Nick cites credibility from prior experience as part of earning trust
- •The first customer’s frustration signaled extreme urgency around RCM
- •Nick rejects some anecdotes about blanket claim denials, but highlights incentive misalignment
- •The real difficulty is the ongoing change in payer rules, not just solving once
- 13:00 – 15:33
Doing things that don’t scale: an API plus intense manual ops and tooling
Nick details the earliest product: a basic API for claim data intake paired with heavy manual review and submission. He describes building reporting in spreadsheets, scaling through internal tooling, and learning the “long tail” of edge cases by doing the billing work himself.
- •Day-one product was an API; the team manually checked and submitted claims
- •Reporting initially lived in Google Sheets and was reviewed weekly with the customer
- •Nick served as the biller for roughly a year-plus to learn edge cases firsthand
- •RCM problems are described as a long tail—“death by a thousand paper cuts”
- 15:33 – 17:55
From services to software to AI: product evolutions and where agents fit
Nick outlines Candid’s major product “lily pads,” moving from selling outcomes (services) to a software platform, then adding front-office workflows. He explains how LLMs/agents are an additive layer, and why Candid prefers upstream prevention over downstream appeals where possible.
- •Early phase sold the outcome because there wasn’t enough product to sell as software
- •Later phases added a software platform and expanded into front-office workflows
- •Agents can be effective for tasks like calling payers, but Candid emphasizes preventing denials
- •Candid combines deterministic rules, ML prediction, agents, and humans depending on the case
- •Token cost and determinism are part of deciding where LLMs make sense
- 17:55 – 19:25
Founder origin story: Palantir experience and choosing RCM as the “big, important” problem
Nick shares how he and his co-founder met at Palantir and what he learned there about operating and execution. He describes gaining exposure to claims adjudication through healthcare-related work and deciding to pursue RCM because it was both large and meaningful.
- •Nick and his co-founder met on the same Palantir team
- •He highlights Palantir as a strong environment for learning operations and execution
- •Exposure to payers and adjudication revealed how manual the process is on both sides
- •RCM appealed as a big market and an important, consequential problem
- 19:25 – 20:44
The product-market-fit moment: beating a long-time incumbent in a competitive RFP
Gustaf asks about the moment Nick felt conviction; Nick points to winning a major competitive process against a long-established competitor. He says the customer was excited by automated RCM as the North Star—even though Candid still had to build additional features for them.
- •Nick describes sustained early growth but lingering uncertainty
- •A competitive RFP win against a decades-old incumbent boosted conviction
- •The customer valued automated RCM as the central differentiator
- •Candid committed to building missing features and executed to meet the customer’s needs
- 20:44 – 23:39
Hiring, team-building, and culture as the scaling constraint
Nick explains how Candid hires and what attributes they look for, emphasizing culture as the most important factor. He describes founder involvement in hiring at scale, minimal reliance on functional interviews, and a consistent “life story” interview scorecard used since day one.
- •Culture is described as the most important determinant of company success
- •Founders remain deeply involved in interviewing and hiring decisions
- •Hiring emphasizes traits like horsepower, grit, and introspection over purely functional screens
- •A consistent “life story” interview framework has remained unchanged since the company started
- •Nick frames culture and hiring rate as eventual limits on scaling
- 23:39 – 27:24
Growth mechanics and economics: early customer acquisition, referrals, and what buyers measure
Nick describes how the first customers were sourced through networking and outreach, and how strong performance drove investor and portfolio referrals. He also explains the economics buyers care about—net collection rate and cost to collect—and why Candid prices lower based on automation advantages.
- •Early customer acquisition relied on targeted outreach through LinkedIn and personal connections
- •Referrals emerged when customers’ investors heard about Candid in board contexts
- •Inbound remains meaningful, but outbound has grown as another lever
- •Nick estimates “cost to collect” often lands around 5–10% all-in for many providers
- •Buyers focus on net collection rate and cost to collect; Candid positions itself as improving both
- •Candid aims to operate at price points competitors can’t sustain due to internal automation
- 27:24 – 30:15
Where AI agents matter most: learning hidden payer rules and mining workflow “data exhaust”
Nick argues that RCM hasn’t had a breakout agent-driven company largely because key payer rules aren’t fully public, making data acquisition difficult. He describes using tracked user actions and workflow history to identify automation opportunities and stresses that RCM is logic- and rules-based, requiring careful versioning and testing.
- •A major barrier is that many payer submission rules are not published publicly
- •Learning often requires submitting claims and calling payers to reverse engineer rules
- •Candid tracks actions across many billers and uses that data to suggest automations
- •LLMs can help identify repeatable patterns users do many times per day
- •RCM is described as logic-based rather than document-based, increasing the need for structured rule management
- 30:15 – 37:33
Series D and what comes next: modular automation, instant payments, and a different future for adjudication
Nick discusses raising a Series D and why the partner is strategically useful for introductions in more traditional healthcare markets. He then outlines product directions: modularizing the automation layer to sit atop other billing systems, pursuing instant payments, and imagining a world where payer and provider systems align more closely.
- •Nick announces a $120M Series D led by Sixth Street
- •He emphasizes the value of deep healthcare connections and introductions
- •Candid is detaching its automation layer to work on top of incumbent billing systems
- •He describes “instant payments” as a focus enabled by granular control and rapid detection of rule changes
- •Nick warns about “agent warfare” if payers and providers deploy competing agents
- •He describes a preferred future where payer/provider systems are aligned and adjudication becomes near-instant
- 37:33 – 38:59
The startup lesson Nick wishes he’d learned earlier: build a financial model sooner
Nick reflects on early-stage advice and says he wishes he’d been pushed to build a financial model earlier. He describes the model as a clarifying planning tool that forces concrete goals and backward planning, and he notes investors compare companies through “spreadsheets.”
- •Nick wishes he had built a financial model earlier in the company’s life
- •He frames businesses as “spreadsheets” that drive investor decisions
- •Modeling forces concrete goal-setting and working backward from targets
- •He contrasts this with early-stage culture that can undervalue spending time on financial planning
