Leading Out Loud: AI Technology

 

Pranav Pasricha, BluePond AI: We see AI being able to do a large percentage of claims tasks that involve reviewing and analyzing documents and data. As of today, we are already achieving very high accuracy in tasks like ingesting claims documents (FNOL, incident reports, adjuster reports, bills, etc.), populating claims systems, calculating reserves, etc. We are also starting to see very good effectiveness in matching loss cause with coverage in policy, automating triage, and flagging cases or payments that need human review.

 

Generic “out of the box” AI models are neither effective nor recommended for claims analysis. Tools specifically built for claims management and embedded within the carriers workflow, can hit upwards of 70–80% automation for tasks mentioned above. With a well-designed AI system, even the tasks or cases that are not fully automated, significant time saving and errors and omissions mitigation can be achieved with AI assistance.

 

Joel Raedeke, Crawford & Company: When we think about the use of AI in claims, we always start with the question: What does our client value? Our clients value the ownership adjusters take in claim strategy and claimant rapport. We never want AI to interfere with that.

 

Tasks that are high-volume, rulesbased, and data-heavy are ideal for AI because they free experts to focus on judgment-driven work. For example, AI can accelerate FNOL and routing through document ingestion and triage, surface anomalies that support fraud detection while leaving final decisions to experts, and identify likely clinical intervention needs that humans validate using context and business judgment.

 

At its current stage, AI is highly effective for straightforward processes and repetitive tasks, reducing cycle times dramatically. But its true value emerges when skilled professionals interpret, validate, and act on AI insights, ensuring fairness and accuracy.

 

 

Pranav Pasricha, BluePond AI: We expect that a large percentage of field work, risk assessment, litigation, and settlement management will remain human oriented in the near future.

 

Building systems for field inspections and estimation requires a combination of robotics and AI, and relying on purely virtual tools can also lead to risk of fraud or leakage or omission of important details.

 

Managing claims settlement and litigation strategy often requires a lot more than just data analytics. Nuanced judgment, understanding behaviors and motivations of different parties, and assessing reputational and relationship factors are often a major part of the equation. While claims managers can rely on AI systems for pattern detection, flagging high fraud risk cases, and suggesting rules-based reserves, human oversight and judgement is often warranted for decision-making.

 

Joel Raedeke, Crawford & Company: The heart of claims is expert judgment, empathy, and accountability, all areas where humans excel and AI cannot replicate. Functions best handled by humans include making complex, high stakes decisions that require nuanced reasoning, applying ethical and contextual judgment that considers fairness and customer impact, and building trust through empathy and ownership in sensitive conversations.

 

These roles require deep expertise and high ownership, qualities that define the best claims professionals. AI can inform these decisions, but humans remain the ultimate stewards of trust and outcomes.

 

 

Pranav Pasricha, BluePond AI: We fully expect a large percentage of document ingestion, data entry, numerical analysis, and desk adjusting to be increasingly automated over time, and human oversight requirement coming down to only the most complicated cases in the next 12–24 months.

 

We foresee increasing fraud and claims padding through the use of AI tools (e.g., fake images and invoices), and correspondingly claims fraud detection solutions to gain more prominence.

 

Joel Raedeke, Crawford & Company: The future is about more empowered human operators, where any efficiency gains are a byproduct more than the primary objective. Key advancements include generative AI that speeds up communication and document work, adaptive AI that learns from human feedback, multimodal tools that use text, voice, and images to improve assessment, and IoT and geospatial data that give experts real time insight. The next frontier is empowering claims professionals with AI as a high-powered tool, enabling faster, smarter decisions while preserving human judgment at the core.

 

Pranav Pasricha, BluePond AI: In our interactions, we find the industry acknowledges the vast potential of AI and is quite accepting of AI-enabled solutions. However, given the sensitivity of our profession, adoption is slower than some other sectors of the economy.

 

Joel Raedeke, Crawford & Company: The industry is experimenting aggressively but scaling cautiously. While many carriers have pilots in place, few have fully integrated AI into workflows. The differentiator will be how well organizations equip their people to use AI effectively through training, governance, and cultural adoption. Those who treat AI as a strategic partner for human expertise will lead.