How is AI used in insurance?

The most documented AI use cases in insurance are AI for claims triage and straight through processing, AI for insurance claims fraud detection and AI for photo based damage assessment in insurance claims, across 132 public deployments by 119 organizations in the Blits.ai AI Use Case Library. 30% of public deployments in insurance put the AI in front of customers directly, against 31% across the library. The rest support employees or run inside a back office process.

Data report

State of AI in insurance

What 132 documented AI deployments by 119 organizations in insurance say about who uses the AI, how far it acts on its own and how often the result is proven, set against the whole library.

Computed live from the library on 12 November 2026. By Len Debets.

132

Deployments

119 organizations in 33 countries

30%

Face the customer

The AI works with customers directly; 31% across the library

71%

Agentic

The AI completes work itself; 63% across the library

56%

Disclose an outcome

A measurable result is public; 61% across the library

Who works with the AI

30% of public deployments in insurance put the AI in front of customers directly, against 31% across the library. The rest support employees or run inside a back office process.

Public deployments by who uses the AI

Back office
65
Employee facing
46
Customer facing
40

How far the AI acts on its own

71% of public deployments in insurance belong to use cases where the AI completes work itself (a supervised or autonomous agent, or an agentic workflow), against 63% across the library.

Public deployments by autonomy level

Supervised agent
72
Assist
42
Copilot
31
Autonomous
3

Where the deployments come from

36% of the public deployments in insurance are by organizations in North America. 58% date from 2025 or 2026, against 57% across the library.

Public deployments by region

North America
48
Europe
46
Asia Pacific
20
Global
7
Middle East
4
Latin America
4
Africa
3

How often the result is proven

56% of public deployments in insurance disclose a measurable outcome, against 61% across the library. Of the 81 reported metrics, 44 came from a vendor rather than from the organization itself.

Who made the claim (reported metrics)

Organization
37
Vendor
44

The most documented use cases in insurance

Ranked by the number of insurance organizations with a public deployment.

AI for claims triage and straight through processing12 organizations
AI for insurance claims fraud detection12 organizations
AI for photo based damage assessment in insurance claims12 organizations
AI copilot for insurance pricing and actuarial analysis9 organizations
AI for commercial underwriting submission intake and triage9 organizations
AI copilot for underwriting risk assessment8 organizations
AI agent for first notice of loss claims intake7 organizations
AI agent for first line contact center service6 organizations
AI agent for insurance policy servicing6 organizations
AI assistant for insurance brokers and agents5 organizations

The benchmarks with the most evidence

KPIs with at least three pooled public values from insurance organizations.

Benchmarks in insurance
KPI	Median	Reported range	Data points	Claimed by
Accuracy	95%	80% to 98%	8	4 organization, 4 vendor
Automation rate	79.5%	45% to 100% Not pooled: up to 70%	6plus 1 up to	3 organization, 3 vendor
Cycle time reduction	65%	50% to 95%	6	3 organization, 3 vendor
Handling time reduction	50% vendor reported	20% to 54%	4	4 vendor
Containment rate	65%	30% to 96%	3	2 organization, 1 vendor

Frequently asked questions

How is AI used in insurance?: The most documented AI use cases in insurance are AI for claims triage and straight through processing, AI for insurance claims fraud detection and AI for photo based damage assessment in insurance claims, across 132 public deployments by 119 organizations in the Blits.ai AI Use Case Library. 30% of public deployments in insurance put the AI in front of customers directly, against 31% across the library. The rest support employees or run inside a back office process.
How many AI deployments in insurance use AI agents?: 71% of public deployments in insurance belong to use cases where the AI completes work itself (a supervised or autonomous agent, or an agentic workflow), against 63% across the library.
How often do insurance organizations prove the results of AI?: 56% of public deployments in insurance disclose a measurable outcome, against 61% across the library. Of the 81 reported metrics, 44 came from a vendor rather than from the organization itself.

About this report

This report counts every public deployment by an organization in insurance, whatever the use case, and sets it against all 1921 public deployments in the library. The industry hub counts only the use cases tagged with this industry, so its totals can be lower. Every figure is computed with the rules on the methodology page, from the open data (CC BY 4.0). The sample reflects what organizations choose to publish, which skews toward large organizations and successful projects.