The operational volume exceeds the team's capacity.
Insurance
An AI agent squad for the insurance industry.
Polyant is the framework for building AI agent squads that work inside the insurance industry flows: claims handled end-to-end, IDD checks on proposals before sending, audit inspectable by the compliance team and by IVASS with a standard SQL client.
Polyant puts an AI agent squad to work on the operational flows of the insurance carrier:
- the claim arriving via WhatsApp, email, or customer app is structured in the management system with a time anchor to the notice
- the pre-check of the life or health proposal (POG target market, suitability, pre-contractual disclosure) enters the flow before sending to the customer, even on the intermediary channel
- the periodic reporting to IVASS and the runtime audit registry stay queryable by the head of compliance via a standard SQL client on the carrier's database
03
WHERE IT FITS
When the flow can't hold on its own.
Three typical contexts that recur in insurance carriers.
The adjuster is overwhelmed by multi-channel inbound claims.
The intermediary channel generates complaints that return to the carrier.
Compliance can't reconstruct decisions without the vendor.
The agent squad is composed of vertical patterns calibrated to the regulated vocabulary of the industry and horizontal patterns (audit, compliance, escalation) that work under the same configuration and the same registry.
04 CLAIMS SQUAD
Five coordinated agents for the claim flow.
Handling a claim is not a single step. It is an end-to-end flow where each agent does its part and passes the work to the next, under the same configuration and the same audit registry.
01 FNOL
FNOL Inbound
Receives the claim notice on any channel, extracts the structured data, classifies the claim type, and opens the case in the management system.
See the agent WhatsApp «I had an accident, this is Rossi»
Type Auto liability · medium
Policyholder Rossi Marco
Policy #IT-2024-04471
Case opened · CRM updated
02 TRIAGE
Claims Triage
Classifies severity and processing channel, routes to the right adjuster, and proposes the initial reserve based on similar claims.
See the agent Claim type Auto liability
Severity Medium
Adjuster Dr. Ferrari · Northeast
Initial reserve € 4,200
Routed ✓
03 DISCLOSURE
IDD Gate
Verifies mandatory disclosure on the liquidation proposal: target market, vulnerable customer, POG registry. Blocks or flags in case of anomaly.
See the agent Target market POG COMPLIANT
Vulnerable customer VERIFY
Pre-contractual info COMPLIANT
POG registry UPDATED
1 verification required before send
04 LIQUIDATION
Liquidation Coordinator
Prepares the liquidation proposal, checks the technical reserve, aligns with underwriting for claims out of standard, and sends the adjuster the case summary.
See the agent Auto liability · #2024-0847
Current reserve € 4,200
Proposed reserve € 3,850
Underwriting aligned ✓
Proposal ready · awaiting adjuster
05 AUDIT
Audit Recorder
Records every squad step in the audit registry: rule triggered, data read, decision taken, alternative proposed. The compliance lead queries the registry with a standard SQL client, even months later, for IVASS inspection.
See the agent fnol.receive 09:14:22 ALLOW
triage.classify 09:14:25 ALLOW
idd.check 09:14:31 WARN
liquidation.propose 09:15:02 ALLOW
SELECT * FROM audit_log WHERE claim_id = '2024-0847'
Need other agents for the carrier? We build them together.
Five coordinated agents for the claim flow are an example. Every carrier has different systems, different internal rules, different operational rhythms. The agents that truly serve get built inside the project, based on the work the customer's team does every day.
Agents calibrated to the industry and cross-industry agents that work with them.
The Claims squad (FNOL Inbound, Claims Triage, IDD Gate, Liquidation Coordinator, Audit Recorder) is described in the previous section. Below are the industry agents that join it or live outside the claims flow.
IDD Gate
Suitability pre-check of the insurance product against the customer needs analysis before offer submission. Decision traced for IDD and Reg. 40 audit.
Bordereau Reconciliation
Bordereau reconciliation between carrier, MGA, and broker. Detection of anomalies on premiums and claims, alert on significant deviations.
Compliance Scan
Cross-industry agent for the pre-screening of regulated contracts and communications before sending. In insurance it works alongside IDD Gate on outbound flows.
A case not listed? We build it together.
These are some example agents typical for the insurance industry. For a specific need, the agents that truly serve get built custom inside the project based on the carrier's management systems, internal rules, and regulatory constraints.
06 REGULATORY MAPPING
What the product does, article by article.
Insurance regulation is wide. Polyant does not promise full compliance: it states what the product does day to day to support the requirements the customer's team has to meet. Below, the references that come up most often in conversations with compliance officers in the industry.
AI Act
AI Act: pricing systems for life and health policies
Pricing systems for life and health policies fall among the high-risk systems under the AI Act. The Underwriting AI with Explainability agent produces reason codes for every pricing proposal, with feature importance and counterfactual: the decision stays with the underwriter, inspectable in the audit registry.
IDD
IDD, Insurance Distribution Directive
The IDD Gate agent verifies target market, POG registry, and vulnerable customer detection on outbound proposals to the customer. The rules are declarative, written by the compliance team, versioned, and validated at agent startup.
IVASS
IVASS regulations
The runtime audit registry can be queried via SQL by the customer's compliance team with a standard reading client, without going through a proprietary console. The inspection export for IVASS is composed of the customer's agents' recorded decisions, with timestamp, rule triggered, data read, decision taken, and alternative proposed.
Solvency II
Solvency II, pricing models and reserves
The Solvency II Validator agent runs periodic backtesting on the pricing models in use. Versioning of prompts, knowledge bases, and declarative rules stays with the customer's team, inside the customer's repository.
GDPR
GDPR and automated decisions
The agent's traced registry preserves the logic of every automated decision, inspectable by the DPO to answer the data subject's requests.
Open transparency for the shared responsibility and AGPL detail →
Four questions from the industry's heads.
Exelab's information security management system is ISO 27001 certified, with coverage on the products built, including Polyant. On the technical perimeter of the product, the runtime audit registry records every system decision in PostgreSQL tables inspectable via standard SQL on the carrier's database; the built-in controls (governance, PII detection, rate limit, tool audit) are part of the runtime in the managed profiles; for AI Act high-risk decisions, the agent produces structured reason codes for every outcome. Exelab operates as an EU vendor, a European technology company with legal seat in Rome, with active vendor qualification on the procurement of a primary European bank and institutional partnerships (HubSpot Elite Solutions Partner, Twilio Gold Partner). The full product trust picture lives at /transparency.
Polyant integrates with the industry's management systems through tools and APIs. For proprietary policy administration systems and industry anti-fraud engines, the integration is delivered during the project by the Exelab team. The logic is the same throughout: the agent talks to the tools, the tools talk to the APIs of the customer's system.
The typical pattern for an insurance Claims squad is 8-12 weeks: discovery, declarative rule configuration, integration with the management system and customer channels, hand-off. The main factor is the depth of IDD rules specific to the carrier and the number of insurance products covered. The effective duration is defined in discovery on the real case.
Operational management of the agents, including rule changes, prompt updates, and configuration evolution, stays with the carrier's team, via the graphical admin panel or directly in TypeScript. On top of that, Exelab provides managed services in three profiles: Core, Advanced, Ultra. The profile detail is on the /managed page.
From a 30-minute conversation to the squad in production.
A 30-45 minute conversation to understand how Polyant for insurance applies to the carrier's case and how long it takes to bring the Claims squad into production.