The Fragmentation Problem
"Point solutions are becoming hard to integrate," said CEO Abhinav Shashank. "The reality is that the provider experience is actually degrading from where it was before if you have different point solutions. Because clinicians now have to switch between Point Solution A to do coding, Point Solution B to do authorization, and then come back to the EHR." Each additional tool that promises to save time adds a context switch, and the cumulative overhead often erases the individual gains.
Innovaccer's response is its "Gravity" platform: a shared data layer that sits underneath all of its AI agents and is continuously trained on real-world healthcare data, from claims denials to edge cases. The platform pulls from a customer's EHR systems, claims data, CRM platforms, and finance and HR systems. Every agent that runs on top of it starts with full institutional context rather than operating in isolation.
The practical result is that agents can hand work to each other. A prior authorization agent doesn't just complete its task and stop — it passes the output directly to a coding and denial management agent, which then routes to the appropriate downstream workflow. In a fragmented system, that handoff requires a human. In Innovaccer's model, it happens automatically.
What the Numbers Show
Several of Innovaccer's health system customers have released early performance data. Risant Health reduced its prior authorization processing time from approximately 45 minutes to under one minute after deploying the company's agentic AI. Prisma Health uses the platform to automatically route high-risk patients to case management, and reports improved accuracy in risk adjustment reimbursement as a result. Banner Health and Franciscan Health have both said they are eliminating hours of manual work per day with the agents in place.
Shashank also points to the platform's compounding learning effect as a structural advantage. Because all agents run on the same shared data infrastructure, each one improves not just from its own interactions but from patterns observed across the entire customer base — something a standalone tool with no shared context cannot replicate.
A New Pricing Model to Match
Innovaccer is shifting away from traditional software licensing toward outcome-based pricing, charging per successful task completed rather than per seat or per usage. The logic is straightforward. "If something is costing you $100 to do from a manual process perspective, we will price it at $20, and therefore we are guaranteeing savings day one with that successful transaction," Shashank said.
That pricing model changes the conversation health system CFOs have to have internally. Instead of approving a software budget and hoping for ROI, they are buying a measurable reduction in cost per transaction from day one. For an industry where AI pilots frequently stall at the proof-of-concept stage because they can't demonstrate hard financial returns, that is a meaningful structural change.
The $250 million commitment will fund new model development, platform expansion, and deeper integration with existing EHR and administrative systems. Whether the investment closes the gap between AI efficiency gains and actual cost reduction — the gap the Peterson Institute identified — will be the real test. The case studies Shashank is promising over the next three years will provide that answer.
This analysis is based on reporting from MedCity News and the Peterson Health Technology Institute.
Image courtesy of César Badilla Miranda and Unsplash.
This article was generated with AI assistance and reviewed for accuracy and quality.
