Ensemble and Cohere announced a partnership to build the healthcare industry's first revenue cycle management (RCM) native large language model (LLM), a custom AI model designed specifically for hospital billing and insurance claim processes rather than general-purpose tasks. Unlike existing AI tools that wrap prompts around generic language models, this custom model will be fine-tuned on real RCM tasks, documented procedures, and deep operator expertise to power AI agents across the entire patient financial journey .

Why Can't Hospitals Just Use ChatGPT for Medical Billing?

Most AI billing tools today rely on general-purpose language models wrapped in heavy prompt engineering, which means the AI has to learn the rules of medical billing every time it processes a claim. This approach raises computing costs, strains the model's reasoning abilities, and ultimately hits an accuracy ceiling when attempting to navigate complex, payer-specific behaviors . Think of it like asking a general knowledge AI to become an expert accountant on the fly, every single time.

The problem is particularly acute in healthcare revenue cycle management, where the stakes are high. Hospitals must navigate complex clinical documentation requirements, understand payer-specific denial patterns, and orchestrate multi-step account resolution processes that fall outside traditional electronic health record (EHR) systems. Generic AI simply isn't powerful enough to compete with the sophisticated denial algorithms that insurance companies use .

How Does a Custom RCM Model Actually Work?

Rather than teaching the AI the rules of medical billing through prompts at inference time, Ensemble and Cohere are embedding that logic into the foundation of the model itself. The LLM will be fine-tuned on real RCM tasks, documented procedures, industry-wide denial patterns, and the proprietary operational expertise Ensemble has gathered managing end-to-end RCM for over 30 health systems nationwide . This means the model learns healthcare revenue cycle logic during training, not during every single query.

Crucially, from a security and compliance standpoint, the training process uses zero identifiable client data or protected health information (PHI). Cohere is relying entirely on synthetic datasets created within a strict, HIPAA-compliant environment . This addresses one of healthcare's biggest concerns: using AI without exposing patient data.

The RCM-native model will be embedded into AI agents that power end-to-end orchestration, from patient intake to account resolution. This deeply integrated approach establishes a truly RCM-native intelligence layer capable of comprehending complex clinical, financial, and regulatory language, and efficiently navigating the multi-step rules and documentation requirements set forth by payers .

Steps to Implement Custom Healthcare AI Solutions

• Assess Your Current Workflow: Identify the specific RCM tasks where generic AI falls short, such as payer portal navigation, denial pattern recognition, or documentation requirement interpretation.
• Partner With Domain Experts: Work with organizations that have deep operational experience in your specific healthcare challenge, not just general AI capabilities.
• Use Synthetic Training Data: Ensure any custom model training uses de-identified or synthetic data that complies with HIPAA and other healthcare regulations.
• Integrate as a Complementary Layer: Position the custom AI model as a complementary intelligence layer alongside existing EHR systems, not as a replacement for multi-million-dollar legacy infrastructure.
• Measure Real-World Performance: Track measurable improvements in efficiency, accuracy, and reliability in revenue cycle management before full deployment.

The most strategic element of this custom LLM is where it sits in the hospital IT stack. Hospital chief information officers do not want to buy another system that competes with their multi-million-dollar EHR platforms. Ensemble and Cohere clearly state this RCM-native model is not an EHR replacement. Instead, it is a complementary intelligence layer that sits alongside the EHR to handle the exact tasks that legacy systems struggle with .

"By pairing Ensemble's deep domain expertise with our secure, enterprise-grade AI capabilities, we can create agents that deliver greater accuracy, consistency, and reliability while meeting the highest standards of privacy and security," noted Aidan Gomez, co-founder and CEO of Cohere.

Aidan Gomez, Co-founder and CEO at Cohere

What Makes This Different From Other Healthcare AI Projects?

The key difference is the implementation-first approach. Rather than building a generic AI and hoping it works for healthcare, Ensemble and Cohere are starting with Ensemble's deep operational experience and data, then fine-tuning the model on real RCM tasks. This is the opposite of the current market approach, where companies take a general-purpose LLM and try to teach it healthcare logic through prompts .

Ensemble has been managing end-to-end revenue cycle services for over a decade, powering financial performance for multiple health systems with award-winning RCM results. This operational expertise is now being baked directly into the AI model's training process, rather than being applied as an afterthought through prompt engineering .

The partnership also addresses a critical pain point in healthcare AI adoption: the need for measurable, reliable performance. By grounding the model in Ensemble's operational knowledge and real RCM data, rather than relying on generic language models, the system will be able to better comprehend the nuances of payer requirements, regulatory details, and multi-step processes that are critical to effective revenue cycle management. This could lead to substantial productivity gains and reduced administrative burden for healthcare providers .

Ensemble and Cohere plan to continue their collaboration to refine and deploy the RCM-native LLM to healthcare providers, with the goal of improving efficiency, accuracy, and reliability in revenue cycle management. In the RCM AI arms race between sophisticated payer denial algorithms and provider collection efforts, hospitals need an AI that speaks the native language of the revenue cycle, not a generic chatbot .