Podcast Summary

Podcast: Becker’s Healthcare Podcast
Episode: AI Powered Revenue Cycle Transformation with Doug Proctor of Candid Health
Date: December 9, 2025
Host: Brian Zimmerman
Guest: Doug Proctor, COO & Co-founder, Candid Health

Episode Overview

This episode explores the transformative role of artificial intelligence (AI), particularly generative models and machine learning, in healthcare revenue cycle management (RCM). Doug Proctor discusses how AI is driving efficiency, reducing costs, and enabling scalable operations within healthcare organizations. He offers concrete examples, addresses the integration challenges, and provides actionable advice for healthcare leaders pursuing AI adoption.

Key Topics & Insights

1. Doug Proctor’s Background and Candid Health’s Platform

Timestamp: [00:27–01:57]

• Doug is COO and co-founder of Candid Health, an end-to-end revenue cycle platform serving hundreds of providers across multiple specialties.
• Candid Health automates claim generation, submission, remit processing, ledgering, and reporting.
• The platform features a configurable rules engine, leveraging AI and machine learning to:
  • Increase provider net collections (e.g., reducing denials)
  • Lower cost to collect by automating manual work
  • Build a scalable technical foundation for ongoing improvement

“What powers what we do is sort of a fully configurable rules engine... we deploy AI and machine learning in very targeted and specific places."
— Doug Proctor [00:40]

2. Current State and Value of AI in RCM

Timestamp: [01:57–06:21]

• Definitions: Doug clarifies that “AI” encompasses generative models, large language models (LLMs), classic machine learning, and automation of business logic.
• Where AI Delivers Value:
  • Low-complexity, high-volume workflows: Automating routine, repetitive tasks accelerates human workflows.
  • Example: Automated eligibility response interpretation—LLMs parse “messy and noisy” eligibility data, identify needed additional checks, and update insurance routing seamlessly.
  • Accelerating Feedback Loops: Candid logs user actions, applies clustering, and LLMs generate new rule recommendations, automating future workflows and insights.
• AI is less about full automation and more about augmenting the human process, especially on tasks that are “detail-oriented, poring through a bunch of data.”

“It’s even a little bit less about fully automating away the workflow... but actually just letting AI agents or large language models operate in a way where they can really accelerate the human workflow.”
— Doug Proctor [03:17]

3. Challenges: Infrastructure and Workflow Integration

Timestamp: [06:21–10:14]

• AI adoption often falters due to outdated workflows and infrastructure that predate AI.
• The crucial starting point is a clear understanding of the problem and the outcome you seek.
• Doug stresses the need for a strong data foundation, robust APIs, and workflow instrumentation to allow AI/ML interventions to be targeted and effective.
• Importance of constraints: Clear definitions of what an AI agent can/cannot do and what data it accesses is key for safe deployment and successful scaling.

“If you’re trying to enter this problem space as a very AI-native, top-down... but you haven’t done the hard plumbing work... it’s really easy to end up in a situation where you’re running an LLM pilot and it looked great on the surface, but when you get into the day-to-day, it’s just not yielding.”
— Doug Proctor [08:31]

4. Mindset Shift and Practical Advice for Leaders

Timestamp: [10:14–12:59]

• Adoption of AI requires a mindset shift—from technology-led to problem-led.
• Doug’s top advice:
  1. Trust Your Intuition:
     • If an AI solution can't be clearly explained—pause. AI shouldn’t be a black box; leaders and users must understand at least the logic, if not the algorithmic detail.
  2. Strategize Short and Long Term:
     • Balance early pilots with long-term scalability planning. Focus on initial metrics, but set the stage for expansion and sustainability.
  3. Be Problem-Oriented:
     • Start with the user and real workflow pain points, not with shiny new tech.
  4. Prepare the Ground:
     • Assess whether your tech stack actually enables both near-term use cases and future scaling to avoid "pilot purgatory."

“If you’ve been doing this work for five, ten, twenty, thirty years... and you don’t understand how it’s actually going to do what they’re saying, you should make sure the conversation doesn’t move forward.”
— Doug Proctor [10:48]

5. Selecting the Right Problems for AI

Timestamp: [13:10–13:51]

• The interconnectedness of RCM problems means careful selection of projects is essential.
• Lead with understanding the user’s day-to-day and the specific challenge before implementing technology.

“Our philosophy is very much like: start with the problem, go get on the ground, in person with the user... and just be smart about the order of operations.”
— Doug Proctor [13:43]

6. Final Reflections & The Future of AI in RCM

Timestamp: [14:08–15:48]

• Doug is “incredibly bullish” about AI’s potential to transform healthcare RCM but emphasizes pairing AI with broader technology modernization.
• AI is not a panacea; value comes when it's integrated with process and broader strategy.
• Healthcare faces escalating cost pressures (rising wages, lower reimbursements)—AI can “move the needle,” but only as part of a holistic approach.

“These are solvable problems. And I personally would say that I’m incredibly bullish on AI’s ability to really move the needle... there actually is a huge amount of opportunity to drive significant impact in ROI.”
— Doug Proctor [14:24]

Notable Quotes & Memorable Moments

• “AI is not going to be some magic box that figures out exactly what you want... without needing any prompting.”
  — Doug Proctor [07:28]

• “If it’s garbage in, it’s garbage out... you’re not going to be able to scale the thing, which is really where the promise lies...”
  — Doug Proctor [09:40]

• “It kind of sucks if you do a lot of work, pilot it, people are pumped, but then you don’t actually have the ability to scale the thing...”
  — Doug Proctor [12:50]

Important Timestamps for Segments

• [00:27] – Doug on his background and Candid’s scope
• [02:36] – Current state and value of AI in RCM
• [03:50] – Detailed example: Automated eligibility response
• [06:21] – Challenges of integrating AI with legacy workflows
• [10:48] – Doug’s advice for leaders: transparency and strategy
• [13:10] – Importance of problem selection
• [14:08] – Doug’s closing thoughts on AI’s future in RCM

Structured Takeaways for Healthcare Leaders

• Define the problem and outcome clearly before launching AI efforts
• Build the right data and workflow foundations before introducing AI/ML
• Don’t accept “magic box” explanations—demand transparency
• Pilot with an eye to future scale, not just quick wins
• Pair AI with broader infrastructure improvements
• Focus on real user problems, not technology for its own sake

This episode offers pragmatic optimism about AI’s potential in healthcare revenue cycle management and a roadmap for leaders seeking real, sustainable value from these technologies.