Automating state controlled-substance reporting: rules for the known, people for the exceptions

1. The task

   Daily reports to every state program, each with its own format
2. In parallel:
   Rules

   State-specific formatting and delivery over a secure pipeline
3. In parallel:
   Security

   Highest-tier health-security classification for PHI transfers
4. Rollout

   Phased state by state
5. Exceptions

   Failures and rule changes routed to a person

Product workflow, shown through the decisions and handoffs.

What I learned

Reporting automation earns trust through explicit requirements and visible exceptions, with people responsible for resolving incomplete work.

From my original account

State reporting automation

I led state controlled-substance reporting automation, coordinating product scope and operational exception handling.

The reporting workflow prepares state-standard dispensing files and delivers them securely. Failures and changing requirements need a clear exception path.

Traceability, reviewable requirements and visible exceptions help make reporting automation dependable.

This was reporting automation, distinct from customer, clinical and billing AI work.

I also launched the customer-care console and personally led duplicate-account search, embedded chat, packet workflows and clinical-ticket routing.

Customer-care contacts often arrive with relevant history that the representative must reconstruct before helping. I envisioned assistance that brought useful context into view sooner and worked with customer-experience and contact-center partners to bring it into the product. Source-linked context mattered: a representative needed to inspect the basis for a suggestion, not take polished text on faith.

Around that core, I worked on self-service assistance, representative support and conversation review. Retrieval from approved knowledge can help with routine questions; suggested answers must remain distinguishable from actions; a human needs sufficient context when a conversation exceeds the assistant’s scope. Product evaluation included quality, latency, cost and correction effort. Clinical authority remained with qualified people.

Other work involved assistance with billing exceptions and prescription data entry. I separated proposing a correction from confirming its effect, and paired automation with quality evaluation and operator feedback. A higher automation rate without its error rate would have been an incomplete product claim.

And a failure worth more than most of the successes. An assistant stopped handing callers to humans. Every component was healthy; every dashboard was green; the handoff between them had broken. We had monitored the nodes and not the seam. The fix added transfer monitoring and a callback fallback. The lesson has taken up permanent residence: in a system made of systems, instrument the joints.

Tools and methods: secure file delivery, state reporting requirements, auditability and exception handling.

Controlled-substance dispensing has to be reported to each state’s prescription monitoring program. At Amazon Pharmacy the process was fully manual: a small team, hours a day, every day of the year. A missed report risked losing the ability to dispense controlled substances. Each state has its own format, cadence and failure behaviour, so the task looked repetitive but was full of exceptions.

I was the product lead. I drove development and the state-by-state rollout, coordinated compliance and legal, wrote the launch announcement and a cost-to-fill savings two-pager, and earlier wrote the vision document for the program.

The decision

Separate three activities that looked alike on a diagram: formatting a report to a state’s rules, delivering it securely, and deciding what to do when a state rejects it or changes its rules. The first two became an automated pipeline with state-specific formatting rules and failure handling over secure file transfer. The third stayed with people, with every failure surfaced rather than retried silently. Because the pipeline moves protected health information to third-party state systems, we earned the highest-tier internal health-security classification before launch. A related requirement set for a small-format pharmacy defined four routing rules so only state-registered, physically in-state pharmacists handle that state’s prescriptions.

A phased state rollout de-risked what was really dozens of separate integrations. Each state went live only when its format had been accepted and its failure path exercised.

About 2,250 reports a day to all fifty states and the District of Columbia, covering all controlled-substance dispensing. About 5 hours a day of manual work eliminated with zero compliance gaps. The four-person, 365-day function was redirected.