PROJECT / 01 · LAUNCHED
KageFlow
An AI receptionist for Canadian plumbing businesses that answers forwarded calls and captures the conversation, and can create customer records, appointments, estimates, or follow-up work when the call outcome requires it.
Dashboard walkthrough15 screens
A synthetic demonstration workspace showing how KageFlow can turn calls into structured operations. Every visible person, business, contact detail, address, and activity record is fictional demonstration data; no customer data appears in this gallery.
SYSTEM MAP / FIVE SURFACES
Choose a component
One product, separated into focused operational boundaries.
ARCHITECTURE / IMPLEMENTATION-ALIGNED VIEW
Two paths, two trust models
The dashboard uses a server proxy and typed API. The voice runtime uses deterministic, business-scoped trusted operations directly against its data service.
Dashboard request path
Voice operation path
Server proxy vs browser credentials
The proxy centralizes routing and configuration without exposing backend credentials in the browser, at the cost of an extra network hop.
Direct voice data operations
Avoiding an API hop lowers call-path overhead, but enlarges the trusted surface and requires deterministic handlers plus business scoping.
Manual release promotion
Immutable images make artifacts reproducible and promotion ownership explicit, at the cost of slower releases than full auto-deploy.
