Platform
Everything between a lead and a claim
Five capabilities share one data model, one event stream and one audit trail.
Why Insurance AI
Intelligent analysis. Controlled decisions. Complete traceability.
Most AI in insurance is a black box that nobody can explain. Here the model never decides: it informs a rule chain you control, and every step is recorded.
- AI advises
01
Intelligent analysis
Language models read medical reports, financial documents and fraud signals, and turn them into scored, explained evidence.
Medical risk 38 / 100
- Rules decide
02
Controlled decisions
A deterministic, versioned rule chain makes the call. It is configurable by your team and behaves identically every time.
Fraud below 15%: pass
- Audit proves
03
Complete traceability
Each input, rule and outcome is written to an immutable trail that can be replayed to explain any decision.
Decision v12, replayable
Seven possible outcomes, each one reachable only through the rules
- Auto Approve
- Approve with Loading
- Human Review
- Request Additional Evidence
- Fraud Investigation
- Postpone
- Decline
Lifecycle
Where each capability fits
- 01
Lead
Agent, portal or copilot adds the customer, family or employer.
- 02
Quote
Every eligible plan is priced instantly with deterministic math.
- 03
Case & gates
E-application, agent report, compliance, first premium, history, medical.
- 04
Underwriting
Parallel medical, financial and fraud scoring, then a rule-based verdict.
- 05
Issuance
Counter-offer, reinsurance, beneficiaries and documents gate issue.
- 06
In force
Free-look, premiums, endorsements and renewals on one state machine.
- 07
Claims
Triage, fraud checks, adjudication, payout and recovery.
Under the hood
One foundation under every capability
The same building blocks serve the portal, the agent app and the copilot.
Multi-tenant and role-aware
Each insurer runs in an isolated tenant with its own plans, rules and users, behind JWT authentication.
One policy state machine
Every status change runs through the same state machine and leaves an immutable event behind.
Three LLM providers
Gemini, OpenAI and Anthropic, with a primary and a fallback chosen at runtime. Credentials are encrypted at rest.
Fraud graph
Relationships between customers, policies and claims are analysed as a graph, so a single application is never judged in isolation.
Built for Pakistan
CNIC handling, PKR pricing, takaful as a distinct product type, and PEP and sanctions screening.
A copilot with confirmation gates
Anything that changes data pauses for a yes or no, enforced in code rather than left to the prompt.
Built to be trusted
AI you can put in front of an auditor.
Insurance AI is a prototype built around the questions a regulated insurer asks first.
Decision traceability
Every status change runs through one state machine and leaves an event behind, so any outcome can be traced back.
Human oversight
Borderline cases go to a person. Human Review and Request Additional Evidence are first-class outcomes.
Explainability
Scores come with reasons, and the rule that produced a verdict is recorded next to it.
Configurable rules
Business rules are versioned data your team can change, not behaviour buried inside a prompt.
Enterprise workflow integration
Multi-tenant and role-aware, with the portal, agent app and copilot working on one set of records.
See the full journey on demo data
We will walk you through a lead becoming a policy and a claim, with the copilot driving where it helps.