A prior-authorization workflow in Dock looks like this: the agent reads the order from Epic or athenahealth, drafts the PA packet against the payer's criteria, writes the draft to a Dock row, the clinician approves, and the submission goes out through Olive AI. If the payer denies, the same row holds the appeal draft, the citations to clinical evidence, and the clinician sign-off. Every step is attributed to an agent or a human, with a timestamp the compliance team can replay.
Epic, athenahealth, and Olive AI stay the systems of record for the raw clinical and payer data. Dock is the system of record for what the agent interpreted from that data. Each Dock row carries epic_order_id, athena_encounter_id, or olive_submission_id, plus the agent identity, the clinician reviewer, the decision, and the timestamp. When the agent needs current state, including a fresh denial letter or an updated criteria document, it re-fetches via API rather than trusting the row it wrote yesterday.
The PA queue surface
| pa_row | patient | service | epic_order_id | olive_submission_id | agent_draft | reviewer | status | payer_decision |
|---|---|---|---|---|---|---|---|---|
| pa_8821 | R. Lin | MRI lumbar w/o contrast | epic_ord_44910 | olv_77231 | drafted by ved-clin-1 against Aetna LCD L34762 | Dr. Patel | approved | approved 2026-05-28 |
| pa_8822 | M. Cruz | Dupilumab 300mg | athena_enc_5582 | olv_77244 | drafted by ved-clin-1, cites NEJM step-therapy exception | Dr. Okafor | submitted | denied 2026-05-27, appeal opened pa_8823 |
| pa_8823 | M. Cruz | Dupilumab appeal | athena_enc_5582 | olv_77251 | appeal letter cites prior failed therapies in chart | Dr. Okafor | approved | pending |
The row is the interpretation. The pointer columns are how anyone, including an auditor, gets back to the source of truth.
One worked appeal
Patient M. Cruz is denied dupilumab on first submission. The agent re-reads the athenahealth encounter, pulls the documented failures of topical steroids and methotrexate, drafts an appeal letter that cites the payer's own step-therapy exception language, and writes the draft to row pa_8823. Dr. Okafor opens the row, reads the cited chart sections inline, edits one paragraph, and approves. The submission to Olive AI is the consent gate: nothing leaves Dock until a clinician's signature is on the row. The payer response comes back to the same row, attributed to the inbound webhook, not silently overwriting the agent's draft.
Why this matters
The AMA reports that 24% of physicians say prior authorization has led to a serious adverse event for a patient in their care, including hospitalization, permanent impairment, or death, and that practices spend roughly 12 hours per physician per week on PA work (AMA 2024 survey). The bottleneck is rarely the first draft. It is the appeal, the re-keying, and the lost paper trail when a denial sits in a fax queue for ten days.
CMS-0057-F now requires impacted payers to send standard prior-authorization decisions within seven calendar days and urgent ones within seventy-two hours (CMS Interoperability and Prior Authorization Final Rule). That clock only helps if the provider side can move at the same speed, with attribution intact. A Dock row that names the agent, the clinician, and the payer turn-by-turn is what makes that clock survive an audit.
Hand-merged PDFs and shared inboxes do not survive that audit. An attributed row does.
See the pillar at Dock for healthcare operations for the rest of the cluster, and Dock for compliance for how the same attribution model carries HIPAA evidence. The pattern under it all is agent identity and agent audit and compliance, with the same reviewer fields that show up in compliance workflows and the same signing primitives in agent identity.
Run your next denial appeal as a Dock row. Attribution, citations, and clinician sign-off in one place.
FAQ
Does Dock replace Epic, athenahealth, or Olive AI? No. Those remain the systems of record for orders, encounters, and payer submissions. Dock holds the agent's interpretation and the human review, linked by ID.
Who is accountable for the appeal letter? The clinician who approves the row. The agent is named as drafter, but the decision attribution sits with the reviewer, with a timestamp.
What happens when payer criteria change mid-appeal? The agent re-fetches the criteria document from the payer or Olive AI on every run. Stale citations in old rows are flagged for re-review rather than reused.
Can compliance replay a specific PA from start to finish? Yes. The row holds every draft, every reviewer signature, and the pointer back to the Epic order and Olive submission, so an auditor can reconstruct the timeline without leaving Dock.
