As of July 2026, out of 65 publicly disclosed agentic AI use cases in the insurance industry 28% are focused on claims management, 20% on internal processes, and 17% on underwriting & pricing. Looking at the outcomes of these use cases, 75% are focused on productivity gains, while only 2% count for revenue uplift. AXA lost the leading position to Allianz, which is now the top insurance company in agentic AI according to Evident Research. Allianz’ Project “Nemo” demonstrates how agentic AI is being applied to claims management in the company to improve efficiency while keeping the human in the loop.
- 28% of 65 disclosed agentic use cases are focused on claims management, while underwriting & pricing only account for 17%.
- 75% of outcomes are productivity gains, while revenue uplift is only 2%.
- Allianz is #1 on Evident’s ranking and its Project Nemo is the concrete claims example. AXA dropped to #2.
Table of contents
Open Table of contents
Agentic AI Use Cases by Application Area and Outcome
Agentic AI capabilities are growing across the insurance industry as a report from “Evident Insights”*) shows. Until July 2026, Insurance companies have disclosed publicly 65 use cases that have been deployed across multiple application areas.
Distribution of 65 use cases across application areas:
Share of 65 disclosed agentic AI use cases by insurance application area. Claims management is the largest application area at 28%, followed by internal process operations at 20% and underwriting and pricing at 17%. Full breakdown: Claims Management 28%, Internal Process Operations 20%, Underwriting & Pricing 17%, Customer Engagement 12%, Engineer & Developer Augmentation 8%, Risk Management 6%, Customer Analytics 5%, Investment & Trading 3%, Other 1%.
| Application Area | % of total |
|---|---|
| Claims Management | 28 |
| Internal Processes & Operations | 20 |
| Underwriting & Pricing | 17 |
| Customer Engagement | 12 |
| Engineer & Developer Augmentation | 8 |
| Risk Management | 6 |
| Customer Analytics | 5 |
| Investment & Trading | 3 |
| Other | 1 |
The outcomes predominantly focus on productivity gains:
Share of 65 disclosed agentic AI use cases by outcome type. Productivity gains dominate at 75% of disclosed use cases, far ahead of customer satisfaction at 12% and risk avoidance at 8%. Full breakdown: Productivity Gains 75%, Customer Satisfaction 12%, Risk Avoidance 8%, Fraud Detection 3%, Revenue Uplift 2%.
| Outcome Type | % of total |
|---|---|
| Productivity Gains | 75 |
| Customer Satisfaction | 12 |
| Risk Avoidance | 8 |
| Fraud Detection | 3 |
| Revenue Uplift | 2 |
Claims operations are currently the main showcase for agentic AI, while profits are driven by underwriting.
Claims is currently the main showcase for agentic AI, with 28% of the 65 disclosed use cases. Underwriting and pricing account for 17% of all use cases, but only 2% of outcomes are revenue uplift. To improve the combined ratio, insurers must leverage agentic AI to automate and optimize their underwriting processes and improve risk selection and pricing. Moreover the usage of AI is shifting from task-based to process and workflow based, where complete processes can be delegated to AI agents - underwriting is a strong next candidate for implementing more agentic AI workflows.
A Concrete Use Case from Claims Management: Allianz’ Project Nemo
With its project Nemo**), Allianz has launched a platform representing an integrated model that orchestrates multiple AI agents to fully process claims related to spoiled food. The process begins with a photo of the damage and ends with its settlement, reducing claims processing and settlement times by 80% - from several days to just a matter of hours. At the same time, the human in the loop principle is maintained, as a human makes the final payout decision.
The agents involved are:
- Planner Agent: Initiates the workflow and simultaneously serves as the orchestrator agent. It acts as the brain of the multi-agent system and functions as the control center for the sub-agents.
- Cyber Agent: Ensures data security throughout the various process steps.
- Coverage Agent: Verifies the insurance coverage and the policyholder’s entitlement.
- Weather Agent: Checks whether a weather event occurred that aligns with the reported claim.
- Fraud Agent: Scans for signs of potential fraud.
- Payout Agent: Calculates the payout amount and reports it back to the Planner Agent.
- Audit Agent: Reviews the entire process once more, generates a summary of all decisions made by the other AI agents, and hands this over to a human claims adjuster who makes the final payment decision.
This use case makes clear that agentic systems can perform multistep processes and by doing so they shift the workload between humans and machines towards the machines, while a human needs to stay in the loop and make the final decision.
These are the Top 10 Leading Insurance Companies in Agentic AI
Evident Research measures leadership across 30 insurance companies by four categories, called pillars:
- Talent: Capability & development
- Innovation: Research, patents, ventures, ecosystem
- Leadership: In public communications and strategy
- Transparency: Of responsible AI activities
| COMPANY | RANK OVERALL | TALENT | INNOVATION | LEADERSHIP | TRANSPARENCY |
|---|---|---|---|---|---|
| Allianz | 1 | 1 | 1 | 7 | 1 |
| AXA | 2 | 2 | 7 | 3 | 2 |
| Manulife | 3 | 6 | 13 | 1 | 4 |
| Zurich | 4 | 7 | 6 | 6 | 3 |
| Liberty Mutual | 5 | 5 | 8 | 10 | 6 |
| Intact Financial | 6 | 13 | 10 | 4 | 17 |
| Travelers | 7 | 9 | 23 | 5 | 7 |
| USAA | 8 | 4 | 11 | 26 | 13 |
| Allstate | 9 | 15 | 4 | 14 | 16 |
| MassMutual | 10 | 16 | 3 | 18 | 12 |
The Near Future of Agentic AI in the Insurance Industry
As investors increasingly expect tangible financial benefits from the use of AI agents alongside efficiency gains, the most successful insurers will be those that move in three directions:
- From task-based AI to agentic applications
- From productivity-focused use cases to AI that improves underwriting, pricing, and claims outcomes
- From siloed solutions to connected workflows
FAQs
Where are insurers actually deploying agentic AI?
Insurers are deploying agentic AI across multiple application areas. 65 use cases split as follows: 28% claims management, 20% internal processes, and 17% underwriting & pricing.
Why is claims so far ahead of underwriting?
Claims management is a repetitive, core insurance process with a high volume, and can benefit significantly from automation and efficiency gains, making it an ideal candidate for agentic AI deployment. Underwriting, on the other hand, starts with a triage process that often requires more nuanced judgment and risk assessment. Implementing automation for underwriting can be more challenging.
What does Allianz Project Nemo actually do?
Project Nemo orchestrates multiple AI agents to process claims related to spoiled food. The agentic system reduces processing time by 80%, from several days to a few hours, while keeping a human in the loop for final payout decisions.
Who ranks highest on Evident’s insurance AI index?
Allianz is currently ranked #1 on Evident’s insurance AI index. AXA follows on rank #2 and Manulife at the third place.
Are these systems fully autonomous?
No, these agentic AI systems are not fully autonomous. While they can handle multistep processes, a human should still make the final decision, especially in critical areas like claims payouts. This follows the human-in-the-loop (HITL) principle, ensuring that a human retains oversight and final authority while AI agents can handle the workload.
Sources:
*) Evident Insights – Insurance Key Findings Report
**) Allianz – When the storm clears, so should the claim queue
Lukas Belka has been working in insurance and reinsurance since 2009. After many years on the business side, he has been extensively focussing on data modeling and analytics since 2020. In 2025 he started working with agentic AI, and today he advocates for the use of CLI-based tools to improve efficiency for the insurance workforce.