Successful GenAI adoption begins with building a unified, context-aware data foundation. Hospitals, diagnostic providers, TPAs, digital health apps and government platforms are at various stages of digitisation, leading to inconsistencies in structure, completeness an interoperability. In an exclusive conversation, Jagdeep Singh Chawla, Chief Delivery Officer, Emids explains to Bhavya Bagga, Business Reporter (Corporate & Leadership) how the triple strategies of data standardisation before modelling, contextual data engineering & governed interoperability ensure AI is not only implemented but embedded into workflows in a way that consistently delivers quantifiable value.
Despite significant investments, studies show that nearly 95% of Gen AI projects fail to deliver ROI. From your perspective, what are the fundamental reasons behind this gap, especially in the healthcare sector?
Healthcare poses unique challenges for GenAI because the underlying data is highly interconnected and workflows are deeply contextual. In markets like the United States, data sits across EHRs, claims systems, labs, imaging platforms and care management tools, all using different standards. This fragmentation makes GenAI training extremely difficult. In India, the fragmentation looks different. Data is distributed across hospitals, diagnostic networks, TPAs, pharmacy systems and government health platforms, with significant variation in how information is captured. While the sources differ from the US, the outcome is similar. AI does not receive consistent or complete context.
A second challenge is how organisations approach AI. Many start with pilots that are not tied to specific business outcomes such as faster prior authorisation, better claims accuracy, reduced manual effort or improved care coordination. Without a defined value target, AI often stays at the experimentation stage. Another barrier is context and explainability. Healthcare workflows require systems that understand clinical logic, regulatory considerations and operational nuance.
GenAI models that rely on pattern recognition alone may perform well in controlled settings but fail when exposed to real clinical or administrative complexity. The gap in GenAI ROI therefore stems less from weak models and more from foundational readiness. Whether in the US or India, organisations need interoperable data, workflow alignment, responsible governance and structured change management to see meaningful results.
How is Emids addressing this Gen AI ROI challenge through its proprietary frameworks or platforms, particularly through innovations like Vibe coding? Could you elaborate on how it differentiates Emids in the healthcare AI space?
At Emids, we have seen that AI delivers value only when it is built around real healthcare workflows rather than as a technical experiment. We have embedded contextual and multidisciplinary engineers working with client organisations to design, refine and scale AI solutions that reflect actual clinical and administrative realities. For example, in the US we work with payer organisations that have complex administrative cycles involving eligibility checks, claims processes and utilisation management. The objective is to ensure AI is grounded in domain logic.
These teams are supported by our healthcare-native contextual AI platform, which incorporates agentic AI patterns and a library of reusable building blocks such as pre- configured agents, workflow orchestrators, audit layers and connectors. These capabilities allow AI to perform supervised multi-step tasks like case classification, data validation, summarisation and escalation rather than simply generating responses.
This model allows us to
- Integrate into existing systems without disruption
- Provide guardrails, observability and governance so AI behaves predictably
- Link each initiative to measurable outcomes such as faster processing, fewer errors, lower
Cost and better patient or member experience
Our differentiation comes from over three decades of healthcare expertise and our ability to embed this domain knowledge into contextual engineering and adaptable AI platforms that support the healthcare ecosystem.
Healthcare data is often fragmented, unstructured, and governed by strict compliance norms. What strategies or technologies do you believe are most effective in unifying and leveraging such diverse datasets for Gen AI success?
Successful GenAI adoption begins with building a unified, context-aware data foundation. In the US, data is often siloed across EHRs, claims engines, pharmacy systems and care management tools, which makes it difficult for AI to understand relationships between clinical decisions and administrative workflows.
India faces similar issues even though the systems are different. Hospitals, diagnostic providers, TPAs, digital health apps and government platforms are at various stages of digitisation, leading to inconsistencies in structure, completeness and interoperability.
Three strategies consistently improve outcomes. The first is data standardisation before modelling. In the US, frameworks such as FHIR and X12 have helped create consistency. In India, ABDM is moving the ecosystem toward structured health data, digital health IDs and provider registries. Aligning with these standards strengthens data quality and improves AI performance.
The second strategy is contextual data engineering. Our AI platform connects data across clinical encounters, claims, care pathways, provider networks and patient or member journeys. This contextualisation allows GenAI to interpret intent and sequence, not just individual data points. The third strategy is governed interoperability.
Responsible data exchange requires cloud- ready architecture with clear access control, lineage tracking and explainability. With these foundations, organisations in both India and the US can move beyond isolated pilots and deploy GenAI solutions that produce consistent and scalable improvements in accuracy, efficiency and experience. Many AI systems in healthcare suffer from learning gaps due to data bias or lack of contextual understanding. How is Emids working to make AI models more clinically relevant and ethically sound? Our approach to AI focuses on making systems transparent, accountable and aligned with clinical and operational workflows. This is essential in both mature digital environments like the US and rapidly modernising settings like India.
We begin with bias-aware data preparation. US datasets often reflect variations in coding practices, payer rules and structured documentation. India adds linguistic and demographic diversity, varied clinical workflows and differences in digital entry quality. Our models are designed to accommodate these nuances through careful curation. We then adopt a clinician and operations driven development approach. Doctors, nurses, administrators, coding teams and care coordinators shape model behaviour so AI aligns with real decision patterns instead of functioning like an unexplainable black box. Continuous drift monitoring ensures models remain reliable as data, patient behaviour or regulatory expectations evolve.
For workflows that influence care quality or claims accuracy, we include human in the loop oversight to ensure responsible decision-making and prevent automation from bypassing essential judgement. We also embed guardrails within our agentic AI framework. These guardrails enforce predictable steps, check for compliance with organisational rules and ensure that AI actions can be explained, audited and corrected when needed. This framework has led to measurable improvements such as faster referral cycles, more accurate classification tasks and reduced administrative errors. Our aim is to ensure that AI decisions are always understandable, traceable and aligned with domain logic.
Why do you see India emerging as a strong testbed for healthcare AI innovation? How do data diversity, regulatory openness, and cost efficiency play into this positioning?
India has become an important environment for healthcare AI innovation because of its mix of scale, diversity and digital maturity. The variation in clinical profiles, demographics and socio-economic conditions helps AI models learn from a wide range of real-world scenarios, which improves their robustness and applicability. India’s digital public infrastructure, including initiatives such as ABDM and India Stack, demonstrates how identity, consent and data exchange frameworks can operate at national scale.
This creates a foundation that supports seamless, secure and interoperable AI-driven healthcare solutions. India also offers deep engineering talent and cost-efficient development cycles, which enable rapid prototyping and refinement. This makes it possible to iterate on AI solutions quickly while maintaining high standards of safety, compliance and operational quality. These advantages position India as a strong testing ground where healthcare AI can evolve at speed and maturity, while still meeting the expectations of trust, compliance and scalability.
Looking ahead, what do you think will define the next phase of ROI-driven AI adoption in healthcare? What role will companies like Emids play in shaping that transition?
The next phase of AI in healthcare will be defined by measurable outcomes rather than model sophistication. Organisations in the US, India and other regions want AI that improves accuracy, reduces manual effort, enhances care coordination and strengthens compliance. Agentic AI will play an important role in this evolution. Unlike traditional models that only generate predictions, agentic AI can perform supervised multi-step tasks within governed workflows.
Organisations are no longer interested in pilots or prototypes. They want AI that clearly moves the needle on
- Administrative speed and accuracy
- Claims integrity and revenue protection
- Member access and care coordination
- Compliance and documentation quality
- Overall operational throughput
If AI cannot demonstrate measurable improvements across these areas, it will not advance beyond experimentation. At Emids, this is exactly where we focus. With our context-led engineering teams working on top of our healthcare-specific contextual AI platform, we design and implement solutions that are anchored to concrete performance metrics from day one.
These metrics may include reducing error rates, accelerating processing times, cutting operational costs or improving experience indicators. Our approach ensures AI is not only implemented but embedded into workflows in a way that consistently delivers quantifiable value. As the industry shifts toward ROI-driven adoption, Emids is helping healthcare organisations move from theoretical potential to practical, scalable impact.
