Prior authorization automation for payers, PBMs, TPAs and providers
At a US health insurer, clinical reviewers spent most of their time on data entry, sorting documents and looking up policy rather than on clinical judgement. Reading each request automatically, checking it against the payer's criteria and drafting the decision takes turnaround from days to hours, and clinicians see only the cases that need them.
What is prior authorization automation?
Prior authorization automation reads each request as it arrives, by fax, portal upload or from the EHR, extracts the clinical facts and checks them against the payer's criteria for that procedure or drug. Requests that clearly meet the criteria can be approved without anyone touching them. The rest reach a clinician with a case summary, the relevant policy excerpts and the supporting evidence already assembled.
- A US health insurer cut average prior auth turnaround from 3.2 days to 4.1 hours, across about 180,000 requests a month.
- The same insurer resolves 70% of requests with no human review, and administrative denials fell from 40% to 6%.
- A US specialty PBM decides specialty-drug requests the same day, down from 6 days, and adjudicates 68% end to end.
- An Indian health TPA decides routine cashless pre-authorisations inside the regulator's 1-hour window, with an adjudicator approving each one.
Prior authorization, step by step
Take in requests on any channel
Requests arrive by fax, provider-portal upload, EHR connection or payer-portal API; at the US insurer, 84% arrived as faxes or PDFs. DocGenie reads every document, classifies it as clinical notes, lab results, imaging report or member ID card, and extracts diagnosis and procedure codes, clinical history, lab values and prior treatments.
Catch missing documents at intake
Gaps are found before anyone reviews the case. At the US insurer, the provider portal now prompts for missing documents before submission, and administrative denials fell from 40% to 6%. At the Indian TPA, missing-document queries go back to the hospital in minutes, which removed the largest single cause of pre-auth delay.
Match the request to criteria
The payer's clinical policies, about 1,400 at the US insurer, are broken into machine-readable criteria: required documents, clinical thresholds, eligibility conditions and exclusions. Rules check eligibility, network status and whether prior auth applies; a model judges whether the documented clinical picture meets medical necessity. The PBM records the criteria version behind each decision.
Decide the clear-cut cases
At the US insurer and the PBM, clear matches are approved automatically; the PBM declines clear failures with the deficiency cited. At the US insurer, a person reviews every adverse decision, including an approval below the level requested, before it reaches the provider: the system drafts, a person signs.
Route the rest to clinicians
Everything else reaches a clinician with the case summarised and the relevant policy excerpts and evidence laid out. At the PBM, reviewers also see a criterion-by-criterion assessment and a recommended decision, which they confirm, modify or override. Every change is logged for audit and for retraining the model.
Clinical judgement stays with clinicians: atypical requests and anything the criteria don't settle. At the US insurer, a person also signs every adverse decision, clinicians check automated drafts in under 3 minutes, and a drafting assistant cut time per case by about 60% on requests that remain human-led. Reviewers' time went to complex cases and care coordination.
Results from 4 deployments
Every figure below comes from the case study it links to.
US Health Insurer
Prior Auth Acceleration
- 70% auto-approved end-to-end
- 180K auth requests / month
- 31% provider abrasion score reduction
US Specialty PBM
Specialty Rx Prior Auth
- 68% end-to-end auto-adjudicated
- 240+ specialty drugs in scope
Health TPA, India
AI CoE for a Health TPA
- 12 use cases in the governed pipeline
- 3 in production in the first quarter
US Healthcare Provider
End-to-End Patient Access Automation
- 24/7 continuous processing
- 4 workflow stages automated
The accelerators behind it
Pre-built accelerators do the work, configured to your documents, rules and systems. Delivery took 12 to 26 weeks in the case studies above.
Prior Auth Accelerator →
Submits, tracks and appeals payer prior authorisations.
Medical Records Parser →
Extracts structured clinical data from PDFs, faxes and scans.
Human-in-Loop Manager →
Approvals, escalations and feedback capture for agents.
DocuMage →
Flagship IDP — OCR + ICR + LLM for any document type, replaces legacy OCR.
Medical Coding Assistant →
ICD-10 / CPT coding co-pilot for revenue cycle teams.
Document processing ROI calculator
Enter your document volumes, handling times and current straight-through rate to see the three-year return, then email yourself the PDF.
Prior authorization: the questions buyers ask
How fast are prior auth decisions once automated?
At a US health insurer, average turnaround fell from 3.2 days to 4.1 hours, and state 24-hour decision requirements are met with room to spare. A US specialty PBM now decides the same day instead of in 6 days, and most cell and gene therapy requests within 24 hours. An Indian TPA decides routine cashless pre-authorisations inside 1 hour.
Will automation deny requests that should be approved?
At the US insurer, the system never sends a denial on its own: every adverse decision, including an approval below the level requested, is reviewed by a person before it goes back to the provider. At the PBM, automatic denials are limited to requests whose evidence clearly fails the drug's criteria, and the specific deficiency is cited.
How many requests need no clinician at all?
At the US health insurer, 70% of requests are resolved with no human review, mostly clear policy matches with clean documentation, such as routine in-network imaging. A further 22% get an automated draft that a clinician approves or edits in under 3 minutes, and only 8% need a full clinical workup. At the specialty PBM, 68% are adjudicated end to end.
Do providers have to change how they submit?
No. At the US insurer, the system takes requests through every channel providers use: the existing provider portal, fax, EHR connections using FHIR prior auth endpoints, and a payer-portal API for third-party submission tools. The portal now also prompts for missing documents before submission, which helped cut administrative denials from 40% to 6%.
Can providers automate their side of prior auth?
Yes. A US healthcare provider automated prior auth as the last of 4 patient-access stages, after scheduling, pre-registration and registration. Bots check coverage, eligibility and network status with the payer and pull clinical notes from the EHR. Prior Auth Accelerator then handles each payer's submission, through the payer's APIs where they exist and its screens where they don't.
Where does patient data stay?
Inside each organisation's own environment. The US insurer and the PBM keep all protected health information, and the model inference, inside a dedicated private network in their own HIPAA-eligible cloud; the insurer ruled out sending any of it to a third-party model API. The Indian TPA runs everything in its own data centre, and no claim document, prompt or model output leaves its network.
How long does it take to deploy?
Delivery took 22 weeks at the US health insurer and 26 weeks at the specialty PBM, which began by digitising its 240-drug criteria library. The Indian TPA took 12 weeks, with 3 use cases in production in the first quarter. The provider's patient-access automation, covering 4 stages, took 18 weeks.
Get the MindMap capability deck
The platform, the accelerator library and how an engagement runs, in one deck. Our team emails it within one business day.
More use cases
Medical records and clinical notes →
99.2% clinical coding accuracy at a hospital group, up from 87%
Healthcare revenue cycle →
14% net patient revenue uplift at a 22-hospital US health system
KYC and account opening →
4 hrs onboarding at a UK challenger bank, down from 5 days
See it running on your own process
A 20-minute walkthrough with the engineers who build these deployments: your documents, your systems, the accelerators running.