AI customer support for life and health insurers works today for policy servicing, billing, claim status, document requests and benefits questions, as long as the platform can act in your systems, log every step, and hand coverage, claims and underwriting decisions to licensed staff. The platforms below all publish insurance or health plan work on their own pages; they differ most in how they price, what security they list publicly, and how much of a case they complete without a person.

Regulators now expect insurers to govern vendor AI as carefully as their own. The NAIC model bulletin on insurers' use of AI had been adopted in 25 jurisdictions by April 2026, and it covers AI systems bought from third parties.

Key takeaways

• The NAIC bulletin reaches your vendors. It asks insurers to keep a written AI Systems (AIS) Program that covers AI "whether developed by the Insurer or a third-party vendor", and suggests contract terms giving audit rights and regulator cooperation.

• Adoption is broad. NAIC's adoption map lists 24 states and the District of Columbia as of April 1, 2026, with separate insurance-specific AI guidance in California, Colorado, New York and Texas.

• Health lines need a BAA; check life lines with counsel. HHS lists health, dental, vision and long-term care insurers among HIPAA health plans. Life insurance is not on that list.

• Separate answering from deciding. NAIC's own guidance says human oversight remains an important part of insurance decision-making.

• Only two of the seven publish a per-outcome price on the pages we checked: Lorikeet and Fin.

How we chose and checked these

We included platforms whose own public pages describe insurance policyholder service, health plan member service, or a named insurer customer, and that can take actions rather than only answer FAQs. Every vendor fact was checked on the vendor's own pages on 23 Oct 2026, and regulatory facts come from NAIC and HHS. Numbering is for navigation, not a ranking. Lorikeet publishes this guide and is one of the tools listed. We have tried to describe every product, including our own, using only what each vendor publishes, and we link to the source for each claim.

The 7 platforms at a glance

1. Salesforce Agentforce: insurers already running on Salesforce

Salesforce's insurance page describes one platform for sales, service, marketing, quoting and claims, with policy administration that includes "claims management, and group benefits management", and Agentforce for Financial Services for 24/7 customer service. On the health side, its Health Cloud page says Agentforce for Healthcare includes a benefits verification skill, and that its contact center tools give staff eligibility, benefits and claims status in one view.

Best for: carriers whose policy, CRM and service data already live in Salesforce, where the AI can read the same records staff do. Trade-offs: the insurance page is titled for property and casualty, so life and health buyers should confirm which industry data models apply to their lines. Much of the published material is about helping producers and staff, alongside customer-facing agents. Pricing for these agents was not published on the pages we checked.

2. Lorikeet: multi-step servicing with deterministic regulated steps

Lorikeet is an AI customer support platform for insurers, MGAs and insurtechs. Its insurance page says it resolves first notice of loss intake, claims status, policy changes, billing and coverage questions end to end across chat, email, voice and SMS, and hands underwriting or sensitive claims decisions to your team with the full conversation. Its healthcare page adds benefits verification and prior authorizations, and its Trust page states SOC 2 Type II, ISO 27001:2022, HIPAA compliance and signed BAAs, with zero-data-retention agreements with model vendors.

The closest insurance proof is a property insurer: SageSure, which writes home insurance in 16 states, resolves alarm certificates and cancellations end to end by email, while premium increase and refund timing questions go to a person.

Limits: there is no published life or health insurer story yet, Lorikeet does not hold PCI DSS, so premium card capture should stay in your payment flow, and voice is available in the US, UK and Australia. Pricing: Start is $2,100 a month and Scale $5,100 a month, paid annually, plus $0.99 or $0.90 per resolved chat, email or SMS ticket; escalations are not charged.

3. Sierra: health plan member journeys

Sierra's healthcare page lists work that maps closely to health insurers: authentication and issue-based routing to replace IVR, guided enrollment with eligibility verification, proactive benefits guidance, claim status, EOBs, appeals and follow-ups. One example on the page is a Medicare member calling to confirm physical therapy coverage and her costs. The page also describes a partnership with Stellarus to help health plans, naming Blue Shield of California.

On the insurance side, Sierra's Marshmallow story describes a UK car insurer's agent handling cancellations, policy updates and renewals, built with Sierra's Agent SDK; customer satisfaction on AI-handled conversations exceeds 82%, while human agents still score higher overall.

Best for: large health plans that want a partner to build member journeys, and teams with engineers who want to own the agent through an SDK. Trade-offs: the healthcare page says it meets required security and clinical standards but does not list specific attestations, and pricing is not published.

4. Parloa: voice-first policy servicing

Parloa's insurance page is built around phone conversations. It describes AI agents capturing first notice of loss, resolving premium questions and processing payments "within PCI-compliant guardrails", and updating address, beneficiary, bank details and marital status in one conversation with a full audit trail. It also describes voice-based identity verification using policy number, date of birth and dynamic challenges, and scaling during storms or catastrophes.

Best for: insurers whose volume is mostly inbound calls; the insurer quoted on the page is BarmeniaGothaer. Beneficiary updates are a common life insurance request, and Parloa names them directly. Trade-offs: the page leads with voice, so check how much of chat and email servicing it covers for you; it states high security standards without listing certifications on that page, and pricing is not published.

5. NiCE Cognigy: large-scale IVR and caller verification

NiCE Cognigy's insurance page describes AI agents for claims support, 24/7 service on any channel and handover to a person when needed, plus peak handling during hurricanes and other events. Its featured case study is a Fortune 500 global insurer that first launched conversational IVR, then automated caller identification and verification; the page says automating the IVR and ID&V reduced average handling time by 15%.

Best for: large carriers with high call volume and an existing contact center estate, where the first goal is natural-language routing and verification before a person picks up. The page also notes a Forrester Wave 2026 leader placement for conversational AI. Trade-offs: its published insurance results are about routing and handle time rather than cases resolved without a person, so ask how far it completes servicing work end to end. Pricing is not published.

6. Decagon: member enrollment and benefits questions

Decagon's health and wellness page describes AI agents that guide members through plan selection, onboarding and benefits questions during high-volume enrollment, and mentions protecting Star Ratings. Its financial services page adds secure account issues with "compliance, validation, and auditability", and both pages describe Agent Operating Procedures written in natural language, simulated conversations before launch and A/B testing across agent versions.

Best for: health plans with sharp enrollment peaks and CX teams that want to edit agent logic in plain language while engineers keep control of integrations. Trade-offs: neither page names an insurer customer or lists certifications, and pricing is not published, so ask for references in your line of business.

7. Fin (now part of Salesforce): published per-outcome price

Fin, which Salesforce completed acquiring on 10 Sep 2026, publishes a financial services page that describes Procedures for multi-step processes such as transaction disputes and fraud claims, configurable ID checks, disclosures and escalation protocols for vulnerable customers, and real-time logging of every input, decision and response. It lists SOC 2, ISO 27001, ISO 42001, ISO 27701 and ISO 27018 and states 99.97% uptime.

Best for: insurers that want a published price and a tool that works on an existing helpdesk; its pricing page lists $0.99 per outcome and says Fin works with any helpdesk, including Salesforce and HubSpot. Trade-offs: its regulated-industry page is written for banking and payments rather than insurance, and while the page mentions HIPAA attestation for organizations handling healthcare data, it does not describe health plan member journeys, so health insurers should confirm BAA terms and use cases directly.

Questions to ask before you pick

The NAIC model bulletin gives a ready-made checklist, because it describes what a regulator can ask you about any AI system, including a vendor's:

• Audit rights: will the contract give us audit rights or audit reports from qualified auditors, as section 4.2 suggests?

• Regulator cooperation: will you cooperate with regulatory inquiries about our use of your product?

• Documentation: can we export, for any conversation, what the AI read, what it did and why it escalated?

• Consumer notice: section 1.9 suggests telling impacted consumers that AI systems are in use. How does the platform disclose that?

• Decisions: which actions can the AI take on its own, and which always go to a licensed person?

• Health lines: will you sign our BAA, and which subcontractors see PHI? HHS's business associate guidance requires BAAs down the subcontractor chain.

Related guides: AI support platforms for insurers, insurance compliance guardrails and AI support for insurers with strict audit needs.

What still needs a human

NAIC's AI topic page says insurers remain responsible for complying with insurance laws when they use AI, and that claims professionals, underwriters and customer service representatives still play an important role in reviewing information and exercising judgment. In support terms:

• Underwriting and coverage decisions, including whether a condition or event is covered.

• Claims decisions and denials, especially death claims, where the person contacting you is often bereaved.

• Beneficiary disputes and authority questions, such as executors, powers of attorney and contested changes.

• Complaints and vulnerable customers, where tone and judgment matter more than speed.

• Compliance sign-off on which tasks are automated, reviewed again whenever a state adopts new AI guidance.

Health insurers should also read our BAA and HIPAA guide for AI support. To test Lorikeet on your own policyholder tickets, start a 30-day free trial.