Podcast

Healthcare AI is Not About Bigger Models, but Smarter Intelligence: Penguin Ai CEO Fawad Butt Explains Why

How Healthcare Organizations Can Build the Right AI Intelligence Layer: Insights from Penguin Ai CEO Fawad Butt

Written By : Market Trends

Artificial intelligence is increasingly becoming an important part of the healthcare industry. From clinical decision support and patient engagement to insurance operations, administration, research, and personalized care, healthcare organizations are exploring ways to integrate AI into their operations. However, as AI investments grow, businesses face an important question: Does healthcare really need the biggest and most powerful AI models to achieve meaningful results?

Healthcare is a particularly complex sector, involving sensitive patient information, regulatory requirements, and significant consequences when systems make mistakes. This makes accuracy, reliability, and data security essential considerations when adopting AI technologies. Rather than focusing exclusively on model size or speed, organizations need to identify the right intelligence for their specific requirements.

Fawad Butt, CEO and Co-Founder of Penguin Ai, joined Priya Dialani on the Analytics Insight Podcast to discuss how healthcare organizations can adopt AI more strategically. The conversation explored specialized AI models, the economics of AI investments, ownership of enterprise intelligence, and the growing role of automation in reducing healthcare's administrative burden. Here are the excerpts from the interview:

Why do Healthcare Organizations Not Necessarily Need the Biggest AI Models?

The focus should not be on finding the newest, fastest, or largest AI model. Instead, healthcare organizations need to identify the model best suited to the task they want to accomplish.

Healthcare requires a high degree of precision because mistakes can have consequences beyond administrative inefficiencies. They can directly affect people's lives and healthcare experiences. Therefore, AI models must be fine-tuned using relevant healthcare data and expertise.

According to Bhat, many widely available models are trained on internet content, books, and third-party information rather than practitioner data from hospitals and healthcare organizations. Building effective healthcare AI requires access to high-quality industry-specific data and healthcare experts who can help refine these models.

The objective should be to develop specialized models that understand specific healthcare workflows, rather than assuming larger general-purpose models will automatically deliver better results.

Where are Healthcare Organizations Over-Investing in AI Today?

One major challenge is the growing spend on AI tokens without clear visibility into the returns they generate. Bhat explained that several large organizations initially gave employees extensive access to AI tools, resulting in steadily increasing token consumption.

However, higher consumption does not necessarily translate into measurable business value. Many AI providers sell access to tokens, leaving organizations to determine how those resources can generate meaningful outcomes.

He believes businesses sometimes choose the biggest technology brands or the most popular AI models without evaluating whether they address their specific operational challenges.

Bhat also highlighted the difference between probabilistic and deterministic systems. Large language models are probabilistic, meaning their responses can vary. However, several healthcare processes require deterministic outcomes, where established protocols and defined criteria must be followed consistently.

Combining these approaches could help organizations build more effective healthcare AI solutions, rather than relying too heavily on large language models.

What does Owning the Intelligence Layer Mean for Healthcare Organizations?

The intelligence layer includes machine learning models, customized large language models, smaller language models, and publicly available AI systems. According to Bhat, this layer could become one of the most valuable assets for organizations as AI increasingly automates workflows and improves operational efficiency.

Healthcare organizations must decide whether to outsource this intelligence entirely to external technology providers or retain control over their proprietary knowledge and processes.

Bhat explained that businesses develop unique operating strategies, expertise, and workflows over time. If organizations transfer these capabilities to external model providers without appropriate control, they risk losing their competitive advantage.

Building ownership requires several elements, including collecting and managing organizational data, establishing AI workbenches, involving subject matter experts in model validation, and creating change-management processes that encourage adoption.

He believes healthcare organizations should work with specialized technology partners when they lack the expertise to build and manage their own intelligence infrastructure.

How Should Healthcare Leaders Balance Closed AI Platforms with Open Approaches?

Healthcare organizations do not necessarily need to build large language models from scratch. Instead, they can fine-tune existing models or distill them into smaller language models designed for specific tasks.

Bhat cited utilization management and prior authorization as examples where specialized models can deliver significant value. He explained that a general-purpose large language model might achieve around 70-75% accuracy in assessing prior authorization requests, while a specialized model trained around a particular payer's or provider's processes could achieve more than 95% accuracy.

Such specialized models can give organizations a strong reason to retain ownership of their intelligence.

However, not every activity requires proprietary AI. Tasks such as basic reasoning, document summarisation, entity extraction, and optical character recognition can often be handled using readily available technologies from large AI companies and cloud providers.

The important consideration is identifying which capabilities create a competitive advantage and investing in ownership where specialized intelligence delivers the greatest return.

What Should Healthcare Organizations Change When Evaluating AI Investments?

Traditionally, technology leaders purchased infrastructure and tools before attempting to build intelligence and business value on top of them. AI is changing this approach by letting organizations buy intelligence they can deploy to address specific problems.

Bhat believes healthcare AI investments should now involve both architectural and business considerations. Decisions should not remain limited to technology departments.

While organizations can purchase foundational infrastructure such as cloud services and computing resources, they should carefully evaluate which higher-level components, including models, intelligence frameworks, and applications, they should develop or own.

He emphasized the importance of bringing business leaders, including CFOs, CMOs, and COOs, into discussions alongside CIOs, CTOs, and CDOs.

This collaborative approach can help organizations connect technology investments to measurable business requirements, rather than purchasing tools without a clearly defined return.

What Will the Future of Healthcare AI Look Like?

Bhat believes healthcare will become one of the most important markets for AI adoption because healthcare services affect individuals, families, and communities throughout their lives.

However, he does not expect a single powerful AI model to solve every healthcare challenge. Instead, the future will likely involve specialized intelligence layers developed for different healthcare requirements by technology companies and organizations within the healthcare ecosystem.

Administrative automation will also play a significant role. Several routine activities, such as making calls, entering information, and transferring details between documents, do not necessarily require human intervention.

AI could automate many of these repetitive processes and reduce the administrative burden on healthcare organizations.

Bhat pointed out that administrative transactions currently account for approximately 20% of overall healthcare spending. He believes AI adoption could potentially reduce this share to 5-6% over the next 10 years.

At the same time, he stressed that clinical and diagnostic AI applications require careful evaluation because mistakes in these areas can have serious consequences.

Ultimately, the future of healthcare AI will depend on how effectively organizations combine specialized intelligence, automation, and appropriate human oversight.

Listen to the full discussion on the Analytics Insight Podcast.

Join our WhatsApp Channel to get the latest news, exclusives and videos on WhatsApp

How Social Media Influences Memecoin Market Activity

DeFi Development Expands Solana Treasury to 2.62 Million SOL as CHAD Dividend Rate Reaches 14%

PEPE Gains 2.41% as Weak Trading Volume Raises Concerns Over Recovery

Crypto Prices Today: Bitcoin Near USD 64,688, Zcash Gains 3.41% in 24 Hours

Top 10 Blockchain Oracle Providers in 2026