The conversation around AI often falls into extremes: adopt AI now or risk falling behind. The momentum around this shift is undeniable. In 2024, global corporate investment in AI reached $252.3 billion1 (compared to $189 in 20232), reflecting both confidence in its potential and urgency to act. But in practice, transformation doesn’t come from grand AI platforms promising to reinvent insurance overnight. It comes from embedding intelligence into the workflows that already define how insurers operate: the routine but critical processes that determine speed, accuracy, and customer experience. AI creates real value not as an abstract strategy, but as a practical accelerator at the point of work. By focusing on workflow-level change, insurers avoid the trap of stalled pilots and endless experimentation. They start small, build confidence, and scale what works. The result is a path to measurable improvements today, with a foundation for broader adoption tomorrow. Here are five workflows where the impact is already becoming clear. 1. First Notice of Loss (FNOL) FNOL can be one of the most complex and inconsistent stages of the claims process. A single submission may include a phone transcript, a handwritten statement, and a blurry photo; all of which an adjuster must review once the claim is assigned. AI can accelerate the process by extracting and classifying unstructured data, identifying claim type and severity, and routing claims to the right handler within minutes. Some systems even flag incomplete reports, prompting follow-up before delays set in. For insurers, piloting AI on FNOL doesn’t mean automating every intake but instead training models on the most common claim categories to reduce manual triage and speed assignment. 2. Reserve Validation and Outcome Prediction Reserves are the financial backbone of claims management, but they often rely on judgment that varies across adjusters and offices. Inconsistencies here ripple through balance sheets and erode regulator confidence. AI models trained on historical claims can predict likely settlement outcomes and highlight when a reserve diverges from expectations. This doesn’t displace adjusters’ expertise; it augments it, offering a data-driven second opinion. A practical starting point is retrospective: compare AI-predicted reserves against actual outcomes on closed claims. That baseline builds trust in the model before integrating reserve alerts into live workflows. 3. Spotting High-Risk Claims Early Some claims are routine, while others are far more likely to escalate into litigation, drag on in treatment, or swell in cost. The challenge is knowing which are which before it’s too late. AI can continuously evaluate open claims, generating dynamic risk scores that adjust as new data arrives. That allows supervisors to reallocate resources proactively, directing specialist attention where it’s needed most. Getting started doesn’t require building a perfect predictor of litigation risk. Even a simple model that flags claims with features historically tied to escalation (injury type, jurisdiction, treatment pattern) can help teams focus their attention earlier. This also includes identifying sleeper claims (those that appear minor at first but later balloon in complexity and cost). These claims often carry hidden risks that aren’t fully uncovered until much later. Early detection helps teams intervene sooner, potentially preventing runaway costs. 4. Detecting Fraud Fraudulent activity remains one of the costliest challenges in insurance. Traditional rules-based systems often miss subtle anomalies, while manual review can’t keep pace with volume. AI is well suited for this kind of pattern recognition. By analyzing patterns in claims history and outcomes, AI can surface suspicious activity and highlight claims deviating from expected trajectories. Early adoption can start with anomaly detection pilots on closed claim data, identifying patterns of fraud missed in real time. 5. Streamlining Claims Communication Beyond analysis and decision-making, claims work involves constant communication: status letters, emails, and internal updates. This administrative burden consumes hours of adjuster time each week. AI-driven drafting tools can generate first-pass versions of these communications, leaving the adjuster to review and approve. The benefit is not only efficiency but consistency: messaging improves, documentation is cleaner, and adjusters reclaim time for higher-value work. A controlled way to begin is with one type of recurring correspondence, like status updates, where templates already exist but personalization takes time. Turning Workflows into Impact The promise of AI in insurance today isn’t about abstract platforms or sweeping reinvention. It’s about embedding intelligence into the daily work that drives outcomes. FNOL, reserves, triage, fraud detection, and communication are just a few areas where workflow-level change is already proving measurable value. Partnerships like Origami Risk and Clara Analytics show how this approach can come to life. Clara’s AI capabilities, from risk scoring to reserve validation, combined with Origami’s structured data and configurable workflows, give insurers a practical way to act on insights. Instead of deploying AI in isolation, carriers can integrate it directly into claims operations: triggering alerts, adjusting workflows, and supporting adjusters in real time. By focusing on workflows, not hype, insurers can capture the immediate gains of AI today while building a foundation for transformation tomorrow. Contact us to learn more about what this looks like for your team. References 1. HAI Standford University. Artificial Intelligence Index report 2025. 2. Wisdom Tree, Christopher Gannatti. Evaluating the AI Megatrend in 2023.