Dr. Deepan Kamaraj, director, data analytics and informatics, UPMC Enterprises
Photo: UPMC Enterprises
Penguin Ai, whose agentic AI platform is focused on finding efficiencies in coding, claims, billing and more, recently launched a strategic data partnership with UPMC Enterprises, the innovation and commercialization arm of healthcare provider and insurer UPMC.
The goal is to co-develop and accelerate the advancement of new healthcare-specific AI models, the companies say.
Through the collaboration, Penguin Ai will access UPMC Enterprises' Ahavi data platform, which supplies secure, de-identified, customized test environments for the validation of new AI models to address critical challenges in healthcare workflows, operations and patient outcomes.
Within Ahavi, Penguin Ai will create and test at least three healthcare systems designed to strengthen the doctor-patient relationship and facilitate timely access to care.
We spoke with Dr. Deepan Kamaraj, director of data analytics and informatics at UPMC Enterprises, who holds a PhD in rehabilitation science and technology, to get a deeper understanding of the work that goes into these types of AI models and the partnership.
Q. What is the challenge that drove the two organizations to work together?
A. Across healthcare, one of the greatest challenges in advancing AI innovation is the time it takes to responsibly test and validate new models. Many AI developers face timelines that stretch for months or even years before they can access usable data.
Complex governance processes, privacy concerns and technical silos often delay progress, even for teams with strong algorithms. The result is a growing backlog of promising models that never reach clinical validation because they can't be tested quickly in a real-world context. Innovation slows not from lack of talent or technology but from an inability to access trusted data fast enough.
UPMC and other health systems see the enormous potential of AI to ease clinician workload, enhance decision-making and improve patient outcomes. Yet realizing that potential requires a new level of agility. Developers need secure, high-quality data pipelines that enable model development and validation in weeks, not years.
At the same time, patient privacy and institutional integrity must remain uncompromised. The challenge, then, is about more than just access, it's about accelerating responsible access to real-world data while preserving ethical and operational safeguards.
That's precisely the problem Ahavi was built to solve. Through its robust infrastructure and governance, Ahavi enables innovators to obtain longitudinal datasets spanning months or years of patient journeys within two to four weeks. This dramatically shortens the time between concept and validation, allowing companies like Penguin Ai to iterate and test AI models in near real time while maintaining compliance.
By combining speed, security and scientific rigor, Ahavi transforms the innovation cycle: What once took months of negotiation and data wrangling can now happen responsibly in weeks, moving the entire industry closer to actionable, trustworthy AI in healthcare.
Q. Please describe the Ahavi data platform and how it works.
A. Ahavi is UPMC Enterprises' secure real-world data and analytics platform designed to power research, AI model development and clinical innovation at scale. It aggregates longitudinal, de-identified patient data from across the UPMC ecosystem into a single, secure environment that meets the highest standards of data governance and privacy protection.
Every record is rigorously de-identified and curated to maintain both compliance and analytical utility.
What sets Ahavi apart is its design philosophy. It's much more than a data warehouse – it's an enablement platform that allows innovators to explore, test and validate ideas in a compliant sandbox environment.
Researchers and developers can run advanced analytics and machine learning pipelines directly within Ahavi's governed environment bringing their code to the data, not the other way around. This "safe compute" model ensures innovation happens responsibly while maintaining patient privacy and institutional trust.
Ahavi's architecture also supports scalable collaboration. From a startup like Penguin Ai to a large-scale pharmaceutical company, the platform provides the same foundation: secure access to harmonized, real-world datasets and the computational infrastructure to derive insights.
In essence, Ahavi transforms how organizations use real-world data from a fragmented, compliance-heavy process into a scalable, ethical innovation pipeline.
Q. Please describe the partnership and what the two organizations hope to accomplish.
A. The partnership between UPMC Enterprises and Penguin Ai was formed around a shared vision: to dramatically accelerate the path from AI concept to clinical validation, responsibly.
Penguin Ai is both refining their current models or developing new AI models and systems, including Patient 360, the ability to provide a snapshot of a patient's medical record, and Enhanced Prior Authorization. Both are designed to help clinicians make faster operational decisions and reduce administrative overhead. These systems rely on large, representative, real-world datasets to train and test their algorithms.
Traditionally, obtaining such data can take months or even years, as teams navigate complex governance reviews, privacy checks and multisystem integrations. That delay can stall innovation and limit the ability to iterate quickly.
Ahavi changes that equation. Acting as a data access acceleration layer, it allows innovators like Penguin Ai to securely access de-identified, longitudinal patient datasets often within two to four weeks, without ever touching live clinical systems.
This compressed timeline enables developers to move from hypothesis to validation at the speed of modern innovation. What previously required extended data-sharing negotiations and custom extract, transform and load pipelines can now happen safely within Ahavi's governed, de-identified environment. The result is a streamlined model development lifecycle that upholds privacy while unlocking speed and scalability.
Together, UPMC Enterprises and Penguin Ai are demonstrating what responsible acceleration looks like in healthcare AI. By combining Ahavi's ability to deliver rapid, compliant access to rich, real-world, de-identified data with Penguin Ai's clinical and technical expertise, the partnership is creating a reproducible blueprint for future collaborations.
The goal goes beyond building better AI faster to redefine how responsible innovation happens, proving that governance and agility can coexist. In doing so, Ahavi positions itself as the critical infrastructure that powers faster, safer and more transparent AI development across the healthcare ecosystem.
Q. Ultimately, how can all of this help transform healthcare through AI? What do you expect the outcomes to be?
A. Penguin Ai's mission is to reduce the $1 trillion annual cost of healthcare administration inefficiencies by leveraging cutting-edge AI systems to streamline healthcare operations, lower costs, and enable better outcomes for patients and providers alike.
Ahavi amplifies that mission by acting as the data access acceleration layer that collapses the traditional AI development timeline from months or years to weeks. By providing governed, de-identified longitudinal datasets, Ahavi enables Penguin Ai to move from base model to validated model in a single development cycle, without sacrificing compliance or clinical fidelity.
That ability to iterate quickly, safely and repeatedly is transformative for an industry where innovation is often constrained by the slow pace of data access.
The short "time-to-insight" window changes the rhythm of AI development. Instead of waiting months for curated datasets, Penguin Ai's data scientists can continuously test and refine algorithms within Ahavi's secure environment, learning from longitudinal patient journeys that reflect real-world complexity.
This agility allows them to build new AI models and refine their current small language models, resulting in newer or improved AI systems. These new agents are developed on living, representative data ensuring each of Penguin Ai's new models evolve in step with real clinical realities. The result is AI models that are accurate and explainable, while also timely and adaptable to healthcare's constant change.
In the broader picture, this partnership signals a new development framework for responsible AI acceleration. Ahavi gives innovators the velocity they need to make meaningful progress, while preserving the governance that healthcare demands.
For Penguin Ai and others that follow, it means a future where developing, validating and deploying trustworthy AI can happen in weeks, not years, making the promise of AI-enabled, equitable and efficient healthcare a near-term reality rather than a distant goal.
Follow Bill's health IT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.
The HIMSS AI & Cybersecurity Virtual Forum is free to attend on Nov. 18. Learn more and register.
WATCH NOW: States take the lead on healthcare AI regulation
