Traditional insurance operations are growing increasingly complex and are under pressure to evolve.
For decades, carriers have relied on predefined workflows. If an underwriter makes decision A, the system routes the task one way. If it's decision B, it goes another. This model created consistency and control, which worked when risks were simpler and data was limited.
Today, that model is being tested. Data flows in from telematics, IoT devices, third-party sources, and customer interactions across multiple channels. The number of variables influencing a single decision has multiplied, and with it, the number of potential pathways.
A single claim or submission now requires evaluating far more inputs, far more quickly, than traditional workflows were designed to handle. Carriers can no longer realistically pre-program every scenario or afford the delays that come with rigid processes.
The structure and consistency of Business Process Management (BPM), which was once a strength, is now becoming a constraint, introducing friction in a system that needs to be far more dynamic. What's emerging is not a replacement for BPM, but its next evolution, introducing real-time decisioning into the flow of work.
AI orchestration — powered by agentic AI — is transforming how work gets done. Instead of relying solely on predefined rules and linear workflows, insurers can introduce systems that evaluate inputs in real time and route tasks accordingly.
In simple terms, processes are no longer just following a script. They're interpreting context and determining what should happen next. Rather than asking, "What rule applies here?" systems can evaluate, "What is happening, and what should we do about it?"
That evaluation then triggers the appropriate workflow whether that's straight-through processing, escalation, or human intervention. Instead of routing every claim above a fixed dollar threshold to a human adjuster, for example, an AI-orchestrated system can assess complexity, fraud signals, and customer history before deciding whether to automate or escalate.
AI orchestration builds on existing BPM investments while introducing flexibility. Some processes, such as compliance, reporting, and standardized operations, will still require predefined workflows. Others, like submission triage, can be more dynamic and event-driven.
But adopting AI orchestration isn't as simple as flipping a switch. The instinct to AI-enable existing workflows can be strong, but it's also where many efforts go wrong. Before insurers can realize the value of more intelligent systems, they need to step back and rethink the processes those systems are built on.
If it's broke, AI can't fix it
As with previous waves of automation, one principle remains unchanged: You can't automate a bad process. This lesson became clear during the rise of robotic process automation (RPA), when many organizations automated inefficient workflows and ended up scaling problems instead of solving them.
The same risk exists with AI. Simply adding AI to a process does not improve it. If workflows are flawed with unclear handoffs, inconsistent data, or redundant steps, AI will only amplify those issues at speed. Consider underwriting. If descriptions vary widely between teams due to inconsistent data inputs, AI won't fix the variables. It will simply make uneven decisions faster.
Before introducing orchestration, insurers need to ask a more fundamental question: should this process be automated at all? That requires evaluating workflows end to end. Where are decisions made? Are they consistent? Is the data reliable? Are there unnecessary steps that can be eliminated? Only after those questions are addressed does it make sense to introduce AI.
Keeping humans in the loop
Insurance is a highly regulated, judgment-driven industry. Coverage decisions, claims outcomes, and pricing determinations carry financial and legal consequences that require oversight. While AI can handle routine evaluation and triage, human expertise remains essential for high-impact decisions.
A human-above-the-loop approach allows AI to manage operational tasks including analyzing data, routing work, and handling routine decisions, while humans provide oversight, handle exceptions, and ensure compliance.
This balance is critical not only for accuracy, but for trust. As regulatory scrutiny increases, insurers must be able to explain how decisions are made, ensure fairness, and demonstrate accountability. Transparency, documentation, and governance are foundational to scaling AI responsibly.
Turning AI insights into action
Over the past several years, the industry has seen no shortage of AI experimentation. Carriers have tested use cases across underwriting, claims, customer service, and fraud detection. But many of these efforts remain siloed delivering insights without driving meaningful operational change.
The challenge is not the AI tools themselves, but the lack of integration into workflows. An AI model that identifies potential fraud but doesn't trigger an investigation delivers limited value. A model that recommends underwriting actions but requires manual follow-up creates friction. Without a mechanism to connect insights to execution, AI remains stuck in pilot mode.
AI orchestration provides that connective tissue. It ensures outputs don't just sit in dashboards. They trigger action. It routes decisions across systems, coordinates tasks between teams, and creates feedback loops that continuously improve performance.
In claims, for example, an orchestrated system could intake first notice of loss, analyze images, validate policy details, and determine severity before assigning the claim to the appropriate path. Straightforward cases can be fast-tracked, while complex claims are escalated to adjusters with full context.
Insurers should start by identifying high-impact use cases where orchestration can deliver clear value such as areas with high volume, measurable outcomes, and well-defined decision points. Claims triage, underwriting intake, and customer service workflows are strong candidates.
From there, the focus should be on building the right foundation: clean, accessible data; well-defined processes; and governance frameworks that ensure transparency and accountability. Technology selection matters, but it should follow — not lead — process design.
Finally, organizations must adopt a mindset of continuous improvement. AI orchestration is not a one-time implementation; it is an evolving capability. As models learn, data expands, and business needs change, workflows must be refined. Flexibility and adaptability are key.
Insurers don't need to abandon the BPM systems and processes they've built over decades. They should extend them with intelligence, adaptability, and stronger integration. Those that take a disciplined approach — rethinking workflows, strengthening data foundations, and embedding human oversight — will move beyond experimentation and achieve real operational impact. As complexity grows, the advantage will shift to insurers that can not only automate processes, but orchestrate them transforming static workflows into dynamic, decision-driven systems built for today's risk environment.
Sachin Kulkarni is Executive Vice President and Head of Commercial & Specialty Insurance and MGA, Americas at Xceedance, a global leader of technology-driven business solutions for the insurance industry. He has more than 20 years of experience in technology and operations across insurance, distribution/supply chain, and other service industries. Prior to Xceedance, Sachin has held roles as global head of IT architecture at Westcon and head of IT strategy for the U.S. and Canada at Marsh.
Opinions expressed here are the author's own.
(Featured image credit: Alexander Supertramp/Shutterstock)
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.