Home / Platform
A safety product on top. A broking engine underneath.
Most insurance apps ask for attention they have not earned. Amanati earns it first — through something a family uses every week — and only then does commercial work.
Three rails
The platform runs on three parallel rails. Keeping them separate is a design decision, not an accident of architecture.
Safety
PilotLocation circles, the Bubble geofence, trip history, elderly-first interaction design, two-wheeler inclusion, SOS, crash and breakdown dispatch into assistance provider networks.
Emergency flows carry no commercial content of any kind. Follow-up is permitted only after the case is closed and the household has rated the service.
Engagement
PilotDriving behaviour capture, the Driving Score, monthly household reporting, the Protection Score, and the concierge that decides when it is appropriate to speak.
This rail generates the data that makes both the safety rail useful and the commercial rail accurate.
Commercial
In buildThe AI Policy Vault, renewal assessment, quick quote across four lines, the digital broker engine, and the omnichannel contact centre that converts what self-serve does not.
What the household experiences
Onboarding and consent, asked in context rather than as a wall at first launch. A daily home screen built around the family, not around insurance. Driving reports that arrive monthly and are worth opening. Assistance that works when it matters.
Insurance appears when there is a reason for it — a renewal approaching, a gap identified, a life event — and every recommendation carries a visible reason code.
What the partner runs
A sales desk with prioritised lead queues driven by renewal dates rather than call lists. Case oversight across assistance events. Configuration controls for brand, panel, pricing and language. Partner reporting against the KPI framework.
In white-label deployments the partner controls the brand, the panel and the assistance network. Amanati runs the engine beneath.
The engine room
Four components do the commercial work. Each is documented in full.
AI Policy Vault
Multi-document upload, extraction and assessment. Every field, clause and conclusion traceable to its source. Extraction confidence held separately from interpretation confidence.
How the vault reads a policyAI Concierge
What it knows, when it speaks, what it will not say, and where the boundary between information and regulated advice sits.
Inside the conciergeDigital broker engine
Identification through to issuance and renewal. A full broking capability, not a comparison widget.
End to endDriving behaviour
Capture, scoring, the monthly report, and the loss ratio chain with every link honestly classified.
The profitability caseArchitecture, in the terms a board needs
Nine layers: device and capture, identity and consent, household data, scoring, decisioning, broking, fulfilment, partner surfaces, and reporting. Twenty-four named components, each with an owner, an external dependency, and a build status.
The concierge and the broking engine are deliberately separate systems. The concierge advises and explains; the broking engine transacts. That separation is what makes the compliance position defensible.
Architecture and integrationDeployment
Federated and configurable. Partners run the platform under their own brand, configured to their market, panel, assistance network, languages and pricing — positioned alongside existing partner tooling rather than in place of it.
See the concierge work on a real policy.
A walkthrough from document upload to a bound renewal, with the commercial mechanic annotated at each step.