Sentinel Signal

Three Demo Flows for Denial-Risk Triage

Source: blog/denial-risk-triage-demo-flows.md

All posts Docs Home Home

Post Content

Three Demo Flows for Denial-Risk Triage

The fastest way to evaluate Sentinel Signal is to start with one narrow wedge: denial-risk triage.

Use the same API surface for prior authorization and reimbursement later, but do not start there. Start with denial review queues, exception routing, and "should a human touch this claim before submission?" decisions.

Flow 1: Instant Trial Key -> First Denial Score

Use this when you want to validate the API directly.

  1. Mint a trial key with POST https://token.sentinelsignal.io/v1/keys/trial
  2. Call POST https://api.sentinelsignal.io/v1/score with workflow=healthcare.denial
  3. Inspect risk_level, confidence, details.denial.distribution_profile, and details.denial.routing_source
  4. Check GET /v1/limits and GET /v1/usage to confirm the key is being tracked cleanly

This is the shortest path from "I saw the product" to "I have a working denial-risk response in my own tooling."

Flow 2: Hosted MCP -> Denial Review in an Agent

Use this when you want an agent runtime to evaluate denials without writing client code first.

  1. Connect the hosted endpoint https://mcp.sentinelsignal.io/mcp
  2. Use Smithery (@sentinelsignal/scoring) or any MCP client that supports Streamable HTTP
  3. Call score_workflow with the same denial payload you would send to /v1/score
  4. Route on the returned risk/confidence fields inside the agent

This is the right demo when the buyer is already experimenting with Claude, Cursor, Windsurf, or a hosted MCP gateway.

Flow 3: Same-Key Upgrade -> Keep Running

Use this when the trial proves useful and you want zero migration friction.

  1. Keep the existing trial key
  2. Start Stripe Checkout with POST https://token.sentinelsignal.io/v1/billing/checkout-session
  3. Complete checkout for growth or pro
  4. Continue using the same key on POST /v1/score, GET /v1/limits, and GET /v1/usage

There is no "create a new paid key and rewire every client" step in the MVP. The same key upgrades in place.

What To Show In Every Demo

  • The denial score itself
  • The confidence value
  • The top explanation factors
  • The routing metadata in details.denial
  • The live usage envelope from GET /v1/limits or GET /v1/usage

Why This Wedge First

Denial-risk triage is the easiest story to explain:

  • fewer preventable denials
  • fewer manual touches on low-risk claims
  • cleaner escalation path for high-risk claims
  • direct link between score output and operational action

That makes it a better first entry point than "general workflow scoring."