.jpg)
Outcomes Insurers
Achieve with AgentFX
A shared build framework and governed deployment stack reduce engineering effort and time to production.
New agents go live in days using one shared contract and auto-registration.
Reusable shared modules eliminate duplicated routing, parsing, and integration work.
Per-tenant isolation, audit trails, and quality gates are built in, not added later.
A Unified Build Framework for
Insurance-grade AI Agents
One BaseAgent contract handles routing, retries, parsing, and logging across every agent.
Ready-made agents for fraud scoring, claims lookup, underwriting, risk, and document classification.
OCR, ICR, and five chunking strategies process PDFs, scans, and handwritten documents reliably.
Deterministic fast-path, LLM intent classifier, and safe fallback route every request accurately.
Risk weights, prompts, and LLM provider are adjustable from the admin UI without code.
Every build runs a retrieval gate; nightly evals score groundedness against a fixed bar.
From a Single File to a Governed Production Agent
AgentFX structures every stage of agent development - from build to continuous improvement - in one platform.
One Platform. Every Stage of the Agent Lifecycle
Every agent built on a single contract, reused across every capability.
Single BaseAgent contract
Auto-registration on deploy
Multi-turn conversation support
Automated gates catch regressions before they reach production.
95% citation accuracy floor
Nightly groundedness evaluation
Permanent regression test cases
Isolation, governance, and audit logging built in at the architecture level.
Per-tenant isolated stack
PII redaction at gateway
Versioned configuration, no drift
.svg){kind=small}
.jpg)
Get answers to common questions
Find solutions to frequently asked questions regarding
Mozart AgentFX
No. Each agent is a single file with one class and one decorator. Once added, it appears automatically in the API, admin UI, and router with no additional plumbing.
Yes. Chunking strategies, retrieval weights, prompts, and LLM provider selection are all adjustable from the admin UI and configuration — no code changes required.
AgentFX supports 100+ LLM providers. The active provider and model can be swapped per agent directly from the settings page.
Each deployment is a fully isolated stack locked to a single customer at install time. A middleware layer rejects any cross-tenant request outright.
Yes. Connecting a new API requires only a base URL, auth config, and route in the configuration file. Secrets are stored at the gateway and never in code.