PUBLISHED ON:
November 13 2025
Agentic Modernization of a Core Insurance Platform
A Southern African insurer partnered with 0to60.AI to modernize its core insurance capabilities without disrupting day-to-day operations. The program delivered two outcomes in parallel:
Agentic integration with existing coreinsurance modules (policy, billing, claims, product/rating, CRM/contact center).
Selective module rewrite (starting withhigh-friction workflows) to embed agentic capabilities directly into the core experience, reducing turnaround times, improving accuracy, and enabling new customer-facing self-service journeys.
The result was a production-ready set of internal and customer-facing AIagents that accelerated quotes and claims, improved operational consistency, and reduced manual rework, delivered on a compressed timeline using 0to60.AI’s delivery accelerators.
The insurer had a mature core platform and well-defined processes, but faced common modernization constraints:
Legacy workflows depended on manualtriage, email handoffs, and policy-admin “work queues.”
Slow change cycles due to tight coupling between UI, workflows, and core logic.
Inconsistent decisioning across branches/agents (underwriting exceptions, claims routing, endorsements).
A growing need for digital self-service that did more than static forms.
The insurer wanted AI agents to improve speed and service quality while maintaining auditability, governance, and compliance.
Integrate agentic capabilities with existingcore modules without replacing the core all at once. Rewrite selected modules where the legacy UX/workflow created high cost-to-serve. Enable customer-facing agents for quotes and claims while also delivering internal agents for operational efficiency. Deliver measurable outcomes quickly via a phased, production-first approach.
0to60.AI implemented an “agentic orchestration layer” that sat above (and alongside) core services:
Connectors to core APIs (policyadmin, billing, claims, product/rating engine), document systems, and communication channels (email/contact center/CRM).
Agent tools (function calling) for read/write actions: create quote, update claim reserve, request documents, generate letters, schedule adjuster, etc.
Governance controls: role-based permissions, human-in-the-loop checkpoints, full audit trails, and policy-based guardrails for sensitive actions. This allowed agents to work with the core, not against it, automating workflows while keeping the system-of-record intact.
For modules where the legacy workflow was the bottleneck, 0to60.AI rewrote “thin slices” of functionality (UI + workflow service) while still using the core as the system-of-record.
The rewritten modules included:
Claims intake + triageworkspace (internal)
Quote-to-bind journey (customer-facing)
Document & evidence collection (both internal and customer-facing)
Each rewritten module embedded agentic assistance “by default” (next-best action, dynamic data collection, automated validations, and guided resolution).
1. Claims Triage & Routing Agent
Reads FNOL intake, policy coverage, priorclaims history, fraud signals, and claim type. Recommends routing (fast-track vs standard vs SIU review), sets initial reserve suggestions, and creates tasks for adjusters. Improves consistency and reduces triage backlog.
2. Underwriting Workbench Agent (Referrals & Exceptions)
Reviews quote context, risk attributes, andunderwriting guidelines. Drafts referral notes, proposes endorsements, and explains premium drivers. Routes exceptions to the right underwriting authority with complete evidence.
3. Policy Servicing Agent (Endorsements & Mid-term Adjustments)
Handles common requests: address changes, beneficiary updates, vehicle changes, additional insured, and schedule updates. Pre-validates required documents and checks policy constraints before writing changes back to core.
4. Reconciliation & Billing Support Agent
Investigate failed payments, mismatch between billing and policy state, or refund requests. Generates customer-ready explanations and internal reconciliation logs.
5. Knowledge & Compliance Agent
Answers staff questions using approved policy wording, internal SOPs, and regulatory guidance. Only cites approved sources; escalates when outside scope.
1. Quote & Bind Agent (Digital Sales Concierge)
Collects risk info conversationally (instead of long forms), validating in real time. Runs rating, presents cover options, explains differences clearly, and supports “what-if” scenarios. Hands off to a human advisor when complexity or compliance triggers require it.
2. Claims First Notice of Loss (FNOL) Agent
Guides customers through FNOL intake (incident details, location, third parties). Requests photos/docs, checks coverage, and provides next steps and service timelines. Creates the claim in core and triggers appointments/work orders if applicable.
3. Claims Status & Next-Step Agent
Provides claim status updates pulled from core plus “what happens next” explanations. Proactively requests missing documents and schedules follow-ups.
4. Customer Document Agent
Explains what documents are needed and why, validates uploads, and confirms receipt. Reduces back-and-forth and incomplete submissions.
“Thin-slice” delivery, not a big-bang programcore
0to60.AI used a delivery model that shipped working capabilities everysprint:
2–3 week discovery: map high-volume workflows, identify automation opportunities, define governance.
Rapid Agile Process build: start with one “hero workflow” (e.g., FNOL → triage → claim creation).
Scale out: add agent tools and modules based on measurable impact.
Hardening: security, DR/BCP patterns, observability, and audit readiness.
Accelerators that reduced time-to-production
Prebuilt agent orchestration patterns (tool calling, routing, escalation, memory boundaries). Reusable connectors/adapters for core insurance APIs and document stores. Standardized evaluation harness foraccuracy, safety, and regression tests. Governance templates: audit logs, approval flows, RBAC policy packs.
To operate safely in a regulated environment, the solution included:
Role-based access to agent actions (read vs write vs approve).
Human-in-the-loop gates for sensitive actions (claim repudiation, payouts, underwriting overrides).
Full audit trails: what the agent saw, what it decided, what it changed, and why.
Data boundaries: least-privilege access and tenant/data residency alignment where required. Prompt and policy controls: approved language for customer communications.
While exact metrics varied by line of business, the program delivered measurable improvements in:
Faster quote completion and higher quote-to-bind conversion from a lower-friction journey. Shorter FNOL intake time and reduced rework from missing/incorrect information. Reduced claims triage backlog and improve drouting accuracy Increased agent productivity (internal staff handling more cases per day). Better customer experience via real-time updates and fewer handoffs
Coexistence strategy: integrate with the existing core first, rewrite only where necessary.
Workflow-first design: agents were built around real operational queues and SLAs.
Trust & control: governance wasn’t an afterthought—approvals and auditability were built in.
Production discipline: logging, monitoring, and evaluation harnesses prevented “demo-only AI.”
ZeroToSixty, Inc.
2261 Market Street, STE 5959
San Francisco, CA 94114
United States
+1 (415) 818-0260