- Industry
- Healthcare (regional network)
- Engagement length
- 12 weeks
- Milestones delivered
- M1 Map, M2 Build & Deploy, M3 Prove
- Stack
- Intake system + custom AI operator (HIPAA-aware, processor-role DPA)
Discover
We instrumented the prior-auth intake queue and timed every step. Packet assembly (gathering records, matching payer rules, formatting the submission) was the bottleneck, and it scored high on both Leverage Map axes.
M1 · Map
The Milestone Spec defined the packet logic, the payer-rule data contract, the human checkpoint before submission, and the metric: turnaround time versus the network's baseline over a defined window.
Plotted workflows (Readiness, Impact): Exception triage (78, 84) Automate First, the M1 candidate; Carrier comms (44, 72) Prove the Case, graduating to (66, 76) after Milestone 1; POD reconciliation (70, 40) Quick Wins; Detention flagging (36, 30) Park / Revisit.
M2 · Build & Deploy
We deployed the operator behind guardrails, in shadow first. It assembled packets and surfaced anything ambiguous to a human reviewer with the supporting records attached.
M3 · Prove
Turnaround fell 61% against the baseline, and routine packets moved to same-day. We reported against the network's own intake records, denominator and window stated.
−61%
turnaround vs. baseline · measured window
Expand
With packet assembly proven, appeals drafting was re-scored into Automate First as the next milestone.