Spotlight on Jaspaul Saini, Principal Analyst at Celent

Jaspaul Saini is helping insurance leaders navigate one of the most complex shifts the industry has ever faced: the rise of artificial intelligence as a core capability.

As Principal Analyst at Celent, Jaspaul advises C-level executives across the insurance and financial services sectors on how to build strategies around data, transformation, and emerging technologies. With a background that spans both consulting and carrier environments, he offers a rare combination of deep technical knowledge and practical industry insight.

What Jaspaul is focusing on right now?

Jaspaul’s work today is centered on the emergence of Agentic AI—autonomous systems that can act independently in complex environments. These systems have the potential to transform how insurers operate, collaborate, and scale.

Key areas of focus include:

• Agentic and Multi-Agent Systems: Exploring the potential of autonomous AI models to reshape financial services by enabling distributed, adaptive decision-making.
• Responsible AI: Developing frameworks that strike the right balance between autonomy and human oversight—critical in highly regulated industries like insurance.
• Enterprise AI Strategy: Helping insurers move from pilot projects to scalable, embedded AI solutions that span their operations.
• Data Governance: Emphasizing the foundational role of data quality, governance, and architecture in supporting any meaningful AI transformation.

A Deeper Dive into Agentic AI

In his recent article, “AI Agents on Rails”, Jaspaul outlines how agentic systems need to balance autonomy with safety. It is important to implement governance and “human in the loop” design where critical decisions or high-impact actions should trigger a human approval loop.

The article explores:

What are AI Agents on Rails?

AI Agents on Rails = Autonomous Agents + Operational Guardrails

Examples of AI Agents on Rails:

• Insurance Claims Agents: Triage cases based on policy rules but require human approval for denial decisions. 
• Customer Support Agents: Allow AI agent to search a knowledge base, draft emails, escalate cases - but never allow direct modification of customer accounts. 

How do we give AI agents the flexibility to be useful, while keeping them on track toward safe, predictable outcomes?

Recommended methods include:

• Human in the Loop design
• Scope Constraints
• Feedback Loops and Escalations
• Ethical and Compliance Checks

The Future

• Adaptive Rails
• Agent Swarms

More about Jaspaul

Jaspaul Saini is a Principal Analyst in Celent’s North American insurance practice. He has held senior technology and consulting roles at Exavalu, HCL, EY, Capgemini, and Zurich North America. His expertise includes:

• Enterprise data strategy and cloud modernization
• AI and advanced analytics
• Core systems transformation
• Martech technologies

He holds an MSc in IT from the University of Warwick and a Bachelor’s in Electronic Engineering from Birmingham City University.

Explore more from Jaspaul at Celent.com.