Which AI-native services can take over healthcare prior authorization workflows from a human operations team?

Yes. The market has moved beyond “PA software” toward AI agents that can actually execute the workflow—pull records, determine requirements, submit, chase payers, update the EHR, and escalate exceptions. But the vendors differ substantially in how much of the workflow they can take over.

The main AI-native options

The important distinction

There are really three different businesses hiding under “AI prior authorization.”

1. Provider-side autonomous PA operations

This is closest to your question if you have a human team today that does things like:

Check whether PA is required → find payer criteria → gather chart notes → fill forms → submit → monitor → call payer → obtain status → request missing information → update EHR → escalate denials. elsai is explicitly positioning itself around this full operational lifecycle. Its agents monitor EHR/RCM systems, retrieve clinical and eligibility information, check payer rules, build submission packets, track cases through resolution, and route low-confidence cases to humans.

That's materially different from conventional PA software: you're buying an operational agent, rather than another work queue for your coordinators.

2. AI clinical/utilization-management infrastructure

Cohere Health is particularly interesting if the buyer is a health plan or risk-bearing organization. It reports that its AI can approve up to 85% of authorizations in real time, while the remaining cases go to clinicians; it also offers delegated utilization-management operations.

So Cohere can replace substantial manual UM work, but it isn't simply a “virtual PA coordinator.” Its strongest position is clinical decisioning + utilization management infrastructure.

3. AI agents for the ugly phone work

A surprising amount of PA doesn't end when you click “submit.” Someone still has to call the payer, navigate IVRs, sit through hold times, talk to a representative and document the answer.

That's where Infinitus and SuperDial are particularly differentiated.

Infinitus says its agents take over PA follow-up calls and retrieve requirements, status and updates for specialty drugs, procedures and diagnostics.

SuperDial similarly automates outbound payer calls, including IVR navigation and conversations with payer representatives, and can hand an unresolved call to a human with context preserved.

If your goal is actually eliminating the human ops team

I'd think about the stack this way:

EHR/RCM event ↓ AI determines whether PA is required ↓ AI retrieves payer-specific requirements ↓ AI reads chart / extracts evidence ↓ AI identifies missing documentation ↓ AI constructs & submits PA ↓ AI monitors status ↓ AI calls payer when electronic channels fail ↓ AI updates EHR/RCM ↓ Human handles only exceptions / clinical judgment / unusual payer behavior

Today, elsai is one of the clearest vendors explicitly selling that entire agentic-operations proposition. Cohere is stronger where clinical UM/decisioning is central. Infinitus/SuperDial are especially useful as the “phone-agent” layer rather than as the entire PA stack.

One caveat: I would not interpret “90% automated” vendor claims as “90% of your employees can disappear.” Vendors measure different denominators—calls, requests, approvals, individual tasks, etc.—and the residual cases are often disproportionately difficult. For example, Cohere explicitly says its remaining cases go to clinical review.

If you're evaluating this as a build-vs-buy / vendor landscape question, I can also map these companies across the PA workflow—intake, clinical extraction, eligibility, submission, payer portals, phone calls, status, appeals, and human exception handling—and identify which ones are genuinely capable of running the whole operation rather than just automating one step.

AI-native and intelligent automation platforms are increasingly capable of absorbing end-to-end prior authorization (PA) workflows—shifting the burden away from human operations teams . These systems utilize a mix of large language models (LLMs), optical character recognition (OCR), electronic health record (EHR) integrations, and real-time FHIR API connectivity (accelerated by regulatory shifts like the CMS-0057-F mandate).

The leading AI-driven services that target the complete or near-complete replacement of manual PA operational teams are categorized below by their primary market focus:

Provider-Side & Enterprise Health System Solutions

• Cohere Health : Focuses on clinically-trained, "glass-box" AI that automates complex medical-necessity decisions. It parses unstructured clinical notes and patient charts, applies strict national or health-plan medical policies, and manages end-to-end authorization pathways, auto-approving a high percentage of cases without human touch.
• Humata Health : Deeply embeds intelligent automation into EHR workflows to assemble clinical documentation, evaluate gaps against payer rules, and automate the submission and tracking lifecycle to remove administrative friction.
• Honey Health : Deploys autonomous AI agents specifically built to handle the medical-authorization "long tail" for specialty practices and Management Services Organizations (MSOs), integrating directly with major EHRs like Epic.
• Insight Health : A physician-founded, provider-side AI platform designed to completely take over the heavy clinical documentation matching and submission burden required for approvals.
• Myndshft : Provides a real-time, automated verification and prior authorization platform spanning both medical and pharmacy benefits, utilizing AI to cut down manual queue handling.

Payer-Side & Network-Wide Engines

• Availity AuthAI : Operating across a vast nationwide network connecting millions of providers and hundreds of payers, AuthAI applies health plan medical policies to clinical data to render automated authorization recommendations in under 90 seconds.
• Banjo Health : Leverages AI and machine learning to automate the adjudication and review side of prior authorizations for payers and Pharmacy Benefit Managers (PBMs), drastically reducing the internal human nurse reviewer workload.

Specialized Workflow & Operations Augmentation

• QuickIntell (QuickAuth / QuickVoice) : Noteworthy for tackling edge cases like phone-based and fax-heavy workflows (which usually trip up standard electronic solutions), its AI voice agents navigate payer phone trees and hold times while closing clinical documentation gaps prior to submission.

If you are evaluating these for a specific implementation, would you like to share whether you are looking from a provider/health system perspective or a payer perspective , and if you need integration with a specific EHR like Epic or Cerner?

A new wave of AI-native platforms and agentic systems is stepping up to replace manual, swivel-chair human operations for healthcare prior authorizations (PAs). Rather than just acting as a template-routing tool, these AI-native services parse unstructured clinical notes from EHRs, map them against distinct payer clinical policies, complete forms, handle interactive voice responses (IVR), wait on hold, and talk to payer representatives.

The prominent AI-native and AI-augmented services capable of heavily automating or taking over these workflows include:

• Humata Health — Born from clinical automation roots (and technology from the legacy Olive AI ecosystem), Humata Health provides an AI-native platform designed to automate prior authorizations for both specialty drugs and medical procedures. It features automated medical policy matching, clinical summarization, and status tracking aimed at achieving touchless submissions.
• Cohere Health — A clinical intelligence platform that utilizes AI to streamline utilization management and prior authorization. It is heavily adopted on both the payer and provider sides to auto-approve routine requests by aligning real-time clinical extraction with evidence-based care paths, cutting manual clinical review times significantly.
• Infinitus — Specializes in conversational and voice AI infrastructure built specifically to call health plans. Instead of a human operations team navigating payer IVR mazes, waiting on hold for 45 minutes, and talking to payer agents, Infinitus’s voice AI agents place outbound calls to check statuses, resolve missing info, and push prior authorizations across the finish line.
• Availity AuthAI — Leverages deep multi-payer connectivity alongside codified medical policy analytics. Utilizing standard FHIR APIs (CRD, DTR, and PAS), AuthAI extracts the precise documentation needed directly from a provider’s EMR workflow and provides transparent, auditable decision support to recommend approvals in near real-time.
• Waystar AltitudeAI — Waystar has evolved its cloud platform to include agentic AI capabilities aimed at making the revenue cycle autonomous. Its features handle proactive clinical justification for prior authorizations and automated clinical appeals without requiring constant human operation.
• CoverMyMeds (with Latent Health integration) — While long known for baseline electronic PA routing, CoverMyMeds integrates agent-driven architectures (via partnerships like Latent Health) to analyze unstructured and structured EHR data via LLMs, drastically reducing manual back-and-forth for medication access.

If you are evaluating these for a specific implementation, would you like to focus on provider-side submission automation (EHR-integrated intake) or payer-side/revenue cycle operations (appeals and voice follow-ups)?

If the goal is to replace a human prior authorization (PA) operations team, the market is splitting into several categories. No single AI-native service fully replaces every PA role today; the closest replacements are agentic AI platforms that can intake requests, gather clinical evidence, submit to payers, chase status, handle payer calls, draft appeals, and escalate exceptions.

A useful way to map the landscape:

Leading AI-native PA automation vendors

Provider-side “AI employee” platforms

• Honey Health — focuses on provider-side end-to-end PA workflows: identifying PA requirements, submitting requests, tracking status, and working across EHRs and payer portals. Honey Health Honey Health Can replace: PA coordinators doing repetitive submission, follow-up, and tracking.
• SkyPoint Cloud (skyAuth) — uses AI agents for PA submission, tracking, denial handling, and workflow visibility. Skypoint Can replace: administrative PA queue work.
• Silna Health — specializes in specialty/therapy workflows. Honey Health Can replace: portions of specialty medication authorization operations.
• Mandolin — applies AI agents to specialty drug access workflows including intake, benefits verification, and prior authorization. Mandolin Can replace: specialty pharmacy access coordinators.

AI voice agents for payer calls

• Infinitus — automates payer phone interactions, a major bottleneck where electronic PA workflows break down. Rx Almanac Can replace: staff who spend hours calling payers.
• SuperDial — focuses on payer calls and follow-ups that require phone-based workflows. SuperDial Can replace: call-center-style PA follow-up work.

Payer-side AI decisioning

These are less about replacing provider PA teams and more about replacing payer utilization management labor.

• Cohere Health — provides AI-powered utilization management and authorization decision support for health plans. Cohere Health Can replace: portions of nurse reviewer workflow, especially straightforward cases.
• Anterior — builds AI clinical review tools for health plans. Honey Health Can replace: manual chart review preparation and review assistance.

If you were trying to eliminate a 20-person PA operations team

A realistic AI-native stack would look like:

1. Intake agent

• Reads orders, referrals, faxes, EHR notes.
• Determines PA requirements.

2. Clinical evidence agent

• Pulls labs, imaging, diagnoses, notes.
• Builds the clinical packet.

3. Submission agent

• Completes payer forms.
• Uploads documentation.

4. Follow-up agent

• Checks portals.
• Calls payers when needed.

5. Appeals agent

• Drafts reconsiderations and medical necessity letters.

6. Exception queue

• Humans handle only unusual cases: - ambiguous medical necessity
• peer-to-peer reviews
• missing documentation
• payer disputes

The vendors closest to a “digital PA operations team” today are probably the provider-side agent platforms (Honey Health, SkyAuth, Silna, Mandolin) combined with voice automation (Infinitus/SuperDial). Payer-side automation (Cohere, Anterior) addresses the other side of the transaction.

For a health system, the buying decision usually comes down to whether you want to replace internal PA staff, outsource the workflow, or automate payer decisioning—those are three different markets.

A new wave of AI-native and cloud-native platforms is shifting prior authorization (PA) from a manual, paper-and-fax-heavy human operation to an autonomous or semi-autonomous workflow. These services leverage Large Language Models (LLMs) trained on clinical data, automated electronic data interchange (EDI), and agentic workflows to interpret medical policies, pull chart data, and draft or finalize determinations.

The primary AI-native services and platforms capable of taking over substantial portions of prior authorization workflows—categorized by whether they serve providers or payers—include the following:

1. For Payers & Health Plans (Automating Adjudication and Intake)

• Anterior (formerly Co:Helm): An AI platform built explicitly for health plans to automate complex clinical administrative work like prior authorization and claims adjudication. Anterior ingests unstructured data (faxes, PDFs, clinical scans) and maps them against specific medical policies using modular AI "Actions," creating an immutable audit trail and auto-completing cases where criteria are met.
• Cohere Health: A leader in clinical intelligence for health plans and risk-bearing providers, Cohere’s platform uses AI-driven clinical decision support to handle end-to-end utilization management. It features high rates of real-time, automated approvals (often bypassing human review entirely for routine requests) while aligning with federal CMS interoperability mandates.

2. For Providers & Health Systems (Automating Submission and Documentation)

• Waystar (AltitudeAI): Waystar embeds cloud-native and agentic AI capabilities directly into its revenue cycle management platform. Its prior authorization tool automates status tracking, patient referral checks, and—via newer agentic capabilities—proactive clinical justification and form-filling by extracting the right data from major EHR systems.
• Availity (AuthAI): Availity serves as a massive digital clearinghouse connecting hundreds of health plans. Its AuthAI solution utilizes analytical AI grounded in codified medical policies to read patient data, assist with Coverage Requirements Discovery (CRD), and automate the documentation assembly needed for smooth electronic submissions without auto-denying or deciding arbitrarily.
• Develop Health: An emerging AI-native platform focused on provider-side form-filling. It uses LLMs to search patient charts for precise clinical thresholds (e.g., BMI requirements or past medication failures) and pre-populates PA submissions with direct citations to supporting visit notes, cutting out hours of manual chart-chasing.

Regulatory and Operational Reality Check

While these platforms drastically reduce human touchpoints, fully "autonomous" operations still operate under human oversight frameworks. Most compliant AI tools use a human-in-the-loop or exception-handling model : clear-cut approvals happen instantly via AI matching, while complex or borderline denials are automatically packaged with highlighted clinical evidence to escalate to a human medical director. Furthermore, platforms are rapidly aligning with CMS interoperability rules (like CMS-0057-F) to ensure FHIR/API standard compliance.

Are you looking to automate prior authorization from the provider side (reducing clinic staff burden and denials) or the payer side (automating medical necessity reviews and approvals)? I can dive deeper into the specific integration requirements for either workflow.

Yes. If by “take over” you mean replace a substantial portion of a human PA operations team—not merely give nurses an AI copilot—the market is now segmented into a few credible categories.

The strongest candidates

1. Cohere Health — closest to “replace the UM team”

Cohere is the most interesting option if you're talking about a health plan's prior-authorization operation rather than a provider's back-office staff.

Its platform combines clinical AI, policy/rules, documentation extraction and workflow automation. Cohere currently claims 85% of PAs can be approved in real time, with the remaining cases routed to clinicians for review. It also offers an in-house operating model, APIs, and delegated UM services.

The important distinction is that Cohere isn't merely automating the submission. It is trying to automate the clinical utilization-management workflow itself—including determining whether requests satisfy applicable policies and preparing the cases that still require human judgment.

For a payer asking, “Can I shrink a 100-person PA operation to 20–30 people?”, Cohere is probably the first company I'd evaluate.

2. Waystar — strongest provider-side end-to-end option

Waystar is attacking PA from the provider/RCM side.

Its Authorization Manager uses AI plus a rules engine to determine whether authorization is required, initiate the authorization, assemble required clinical documentation, submit it and track the process. Waystar says its system checks millions of payer rules and processes tens of millions of authorization transactions annually.

That's particularly compelling for a health system where the human team currently does things like:

• check whether PA is required
• identify payer requirements
• gather documentation
• submit through portals
• monitor status
• chase missing information
• deal with authorization-related scheduling/RCM fallout

Waystar is therefore a better answer than Cohere if your “human operations team” sits inside a hospital or physician organization.

3. Infinitus — the killer app for phone-based PA work

Infinitus is somewhat different and, in some organizations, could eliminate an enormous amount of labor.

Its voice AI agents call payers on the provider's behalf to obtain PA requirements, status, updates and other information. This matters because a supposedly electronic PA frequently degenerates into a phone call to the payer anyway.

Infinitus says its agents handle specialty drugs, procedures and diagnostic tests across numerous specialties and can return the results through an API or its portal.

So if your current workflow looks like:

employee → payer portal → fax → hold music → payer representative → transferred department → status update → employee documents it in EHR Infinitus is unusually well suited to remove the human from that loop.

4. Availity — infrastructure rather than autonomous labor

Availity is important because of its payer connectivity and transaction infrastructure, but I'd categorize it differently.

It's more “digitize and route the PA transaction” than “replace the entire PA operations department with AI agents.”

That distinction matters when evaluating vendors. A system can make electronic submission dramatically easier while still leaving humans responsible for documentation, follow-up, clinical review and exceptions.

The emerging AI-native architecture

The interesting thing in 2026 is that PA automation is starting to look less like a single application and more like an agentic workflow:

EHR/order → determine PA requirement → retrieve payer policy → extract clinical evidence → complete PA → submit → monitor → communicate with payer → handle requests for additional information → escalate only exceptions → document outcome

Different vendors own different portions of that chain.

The current market analysis also reflects this segmentation: Infinitus is particularly strong in hub/call automation, Cohere on payer-side utilization management, and Waystar on provider-side authorization/RCM.

If your objective is actually to eliminate the human team

I'd shortlist them differently depending on who you are:

• Health plan: Cohere first
• Large hospital/health system: Waystar + Infinitus
• Specialty pharmacy: Infinitus + an ePA/RCM platform
• Pharma patient-support hub: Infinitus / Neon-type hub automation
• PA outsourcing/BPO replacement: Cohere is the most interesting, because it can combine software with delegated clinical operations rather than simply selling automation software. Cohere explicitly offers delegated UM services alongside its AI platform.

One caveat: “fully autonomous” is still not the same as “no humans.” The strongest platforms automate administrative and straightforward clinical cases, while keeping genuinely consequential clinical determinations and exceptions under human oversight. Cohere, for example, explicitly says that its remaining cases go to clinicians and that AI isn't used to make care-denial decisions.

If you're evaluating this as a build-vs-buy / vendor-selection question, I can also give you a 10–15 company landscape mapped by workflow step (intake → clinical documentation → submission → payer calls → status → appeals), with which vendors can plausibly replace FTEs and approximate automation levels.

Infinitus says its agents handle specialty drugs, procedures and diagnostic tests across numerous specialties and can return the results through an API or its portal.

So if your current workflow looks like:

employee → payer portal → fax → hold music → payer representative → transferred department → status update → employee documents it in EHR Infinitus is unusually well suited to remove the human from that loop.

4. Availity — infrastructure rather than autonomous labor

Availity is important because of its payer connectivity and transaction infrastructure, but I'd categorize it differently.

It's more “digitize and route the PA transaction” than “replace the entire PA operations department with AI agents.”

That distinction matters when evaluating vendors. A system can make electronic submission dramatically easier while still leaving humans responsible for documentation, follow-up, clinical review and exceptions.

The emerging AI-native architecture

The interesting thing in 2026 is that PA automation is starting to look less like a single application and more like an agentic workflow:

EHR/order → determine PA requirement → retrieve payer policy → extract clinical evidence → complete PA → submit → monitor → communicate with payer → handle requests for additional information → escalate only exceptions → document outcome

Different vendors own different portions of that chain.

The current market analysis also reflects this segmentation: Infinitus is particularly strong in hub/call automation, Cohere on payer-side utilization management, and Waystar on provider-side authorization/RCM.

AI-native prior authorization platforms like Cohere Health, QuickIntell (QuickAuth), and Prosper AI can take over routine health insurance prior authorization (PA) workflows from human operations teams by auto-assembling clinical records, checking payer criteria gaps, submitting requests via electronic portals, and navigating phone trees.

While human teams are still required to review complex edge cases or medical appeals requiring licensed clinical judgment, AI-native services manage the end-to-end mechanical and administrative burden.

Capabilities of AI-Native Prior Authorization Services

How AI Replaces Manual Operations

• Data Ingestion & Aggregation: AI pulls structured and unstructured data (PDFs, clinical notes, lab results) from central repositories (like AWS HealthLake or EHR systems) without manual copying.
• Gap Analysis: The software flags missing requirements (such as missing proof of failed conservative treatments) before the request is formally submitted, dropping initial denial rates.
• Multi-Channel Execution: If an electronic portal is unavailable, specialized voice AI handles the 30%–40% of authorizations locked behind phone trees and interactive voice response (IVR) systems.

If you'd like, let me know:

• Whether you are looking for provider-side (clinician/hospital) or payer-side (insurance) automation
• Your primary EHR system (e.g., Epic, athenahealth)

I can narrow down the best integration fit for your team.

Yes. The market has moved beyond “prior-auth software” toward AI agents that actually execute the workflow—reading clinical records, determining requirements, navigating payer portals, making payer calls, submitting requests, tracking status, and escalating only exceptions.

The important distinction is that no single vendor completely replaces a sophisticated PA operations team across every payer and specialty yet. But several are getting close for defined workflow segments.

The strongest candidates

1. Infinx — closest to an autonomous PA operations team

If your question is “Who could I give the whole PA work queue to?”, Infinx is one of the first companies I'd evaluate.

Its Patient Access Plus platform now exposes separate agents for:

• PA determination
• PA initiation
• PA follow-up
• payer/clearinghouse connectivity
• workflow execution
• human exception handling

Its initiation agent can navigate payer/TPA portals, enter patient/service information, submit requests, and pause when something requires human judgment.

Importantly, Infinx isn't claiming that every case is autonomous. Its model is automation + human-in-the-loop for exceptions, which is much closer to how you'd actually replace an operations team.

My take: probably the most interesting vendor if you want to outsource the workflow, rather than just make your existing staff faster.

---

2. Infinitus — exceptional for the “phone hell” portion

Infinitus is particularly compelling if your PA team spends hours calling payers.

Its voice agents can call payers, navigate IVRs/live agents, collect PA requirements and status, and return the information to the customer's system via API. It explicitly supports specialty drugs, procedures and diagnostic tests.

That's a meaningful distinction from a conventional PA platform:

Instead of giving the human a better interface for making the call, the AI makes the call. Infinitus reports, for example, that one specialty-pharmacy customer saved an average of 20 minutes per PA status call.

My take: if 30–50% of your PA labor is phone calls, I'd put Infinitus near the top of the list.

---

3. Humata Health / R1 — strong on clinical-documentation automation

Humata takes a different approach: use AI to understand the clinical record and payer requirements so that the authorization can be prepared correctly the first time.

Its current offering analyzes documentation against payer policies, identifies gaps likely to cause denials, and can generate answers to payer clinical questions.

There's also an important recent development: R1 announced in August 2026 that it had agreed to acquire Humata Health, integrating its PA technology into R1's broader revenue-cycle platform.

That potentially makes the combination particularly interesting for organizations that want AI + an operating infrastructure layer, rather than another point solution.

---

4. Cohere — arguably the most autonomous, but from the payer side

Cohere Health is a slightly different beast.

It is primarily a payer/utilization-management platform, rather than a vendor you hire to run your hospital's PA department.

Its platform automates intake, clinical-policy matching, medical-necessity review and authorization decisions. Cohere says its AI handles the majority of PA decisions and reports up to 85% real-time approvals in some specialties.

It also supports fax/phone/EHR intake and automated extraction of clinical information.

So if you're a health plan asking “Can AI replace a chunk of my UM operation?”, Cohere is one of the clearest answers.

If you're a health system asking “Can AI replace my PA staff?”, it's less directly comparable.

---

5. Rhyme — the connectivity layer

Rhyme is important because a huge part of PA automation isn't intelligence—it's getting the authorization into the payer's system.

Rhyme's EHR-integrated network can connect provider workflows to payers, and it has been deployed alongside Cohere to create touchless authorization workflows. In one Medical Mutual deployment, the combination was positioned as an end-to-end automated PA process.

I'd think of Rhyme less as “your AI PA department” and more as the plumbing that lets an AI PA department actually execute transactions.

---

6. Notable — excellent AI worker, but not full replacement

Notable's Sidekick is more explicitly human-in-the-loop.

It can pull together chart information, summarize documentation, draft payer responses and operate within EHR/payer-portal workflows. Notable says PA submissions that previously took 20–30 minutes can become roughly 7-minute workflows.

That's impressive productivity automation, but I would not put it in the same “replace the operations team” bucket as an autonomous workflow orchestrator.

How I'd segment the market

If you're evaluating these for an actual PA operations replacement, I'd frame the stack this way:

Clinical/document intelligence → Humata / Notable / Cohere

Authorization transaction execution → Infinx / Rhyme

Payer phone calls → Infinitus

Payer-side clinical decisioning → Cohere

Full workflow orchestration + human exceptions → Infinx is particularly interesting

The more interesting architecture is therefore not necessarily picking one vendor. It is an AI PA workforce:

EHR/order → determine whether PA required → retrieve chart evidence → determine payer requirements → complete forms → submit → call payer when necessary → monitor status → respond to requests → escalate clinical exceptions → update EHR → close case The vendors that can execute all of those steps, rather than merely assist a human with one of them, are the ones I'd consider true replacements for an operations team.

My 2026 shortlist

If I were running an RFP, I'd start with:

1. Infinx — strongest candidate for broad provider-side autonomous PA operations.
2. Infinitus — strongest for payer-call-heavy workflows.
3. Humata/R1 — strongest combination of clinical AI + broader RCM operating infrastructure.
4. Rhyme — important if payer/EHR connectivity is the bottleneck.
5. Cohere — especially if you're evaluating this from the health-plan/UM side.
6. Notable — excellent if you want substantial staff reduction but still expect humans to own complex cases.

One caveat: “AI-native” does not yet mean “zero humans.” The credible platforms are generally designed so that AI handles the high-volume deterministic work and humans receive the exceptions involving missing information, unusual payer behavior, clinical judgment, or novel workflows. Infinx explicitly describes this model, while Notable describes itself as human-in-the-loop.

If you're looking at this as a startup/investment landscape, I can also map these companies by which exact PA tasks they automate, customer (provider/payer/pharma), degree of autonomy, funding/ownership, and likely defensibility.

Infinitus is particularly compelling if your PA team spends hours calling payers.

Its voice agents can call payers, navigate IVRs/live agents, collect PA requirements and status, and return the information to the customer's system via API. It explicitly supports specialty drugs, procedures and diagnostic tests.

That's a meaningful distinction from a conventional PA platform:

Instead of giving the human a better interface for making the call, the AI makes the call. Infinitus reports, for example, that one specialty-pharmacy customer saved an average of 20 minutes per PA status call.

My take: if 30–50% of your PA labor is phone calls, I'd put Infinitus near the top of the list.

---

3. Humata Health / R1 — strong on clinical-documentation automation

Humata takes a different approach: use AI to understand the clinical record and payer requirements so that the authorization can be prepared correctly the first time.

Its current offering analyzes documentation against payer policies, identifies gaps likely to cause denials, and can generate answers to payer clinical questions.

There's also an important recent development: R1 announced in August 2026 that it had agreed to acquire Humata Health, integrating its PA technology into R1's broader revenue-cycle platform.

That potentially makes the combination particularly interesting for organizations that want AI + an operating infrastructure layer, rather than another point solution.

---

4. Cohere — arguably the most autonomous, but from the payer side

Cohere Health is a slightly different beast.

It is primarily a payer/utilization-management platform, rather than a vendor you hire to run your hospital's PA department.

Its platform automates intake, clinical-policy matching, medical-necessity review and authorization decisions. Cohere says its AI handles the majority of PA decisions and reports up to 85% real-time approvals in some specialties.

It also supports fax/phone/EHR intake and automated extraction of clinical information.

So if you're a health plan asking “Can AI replace a chunk of my UM operation?”, Cohere is one of the clearest answers.

If you're a health system asking “Can AI replace my PA staff?”, it's less directly comparable.

---

5. Rhyme — the connectivity layer

Rhyme is important because a huge part of PA automation isn't intelligence—it's getting the authorization into the payer's system.

Rhyme's EHR-integrated network can connect provider workflows to payers, and it has been deployed alongside Cohere to create touchless authorization workflows. In one Medical Mutual deployment, the combination was positioned as an end-to-end automated PA process.

I'd think of Rhyme less as “your AI PA department” and more as the plumbing that lets an AI PA department actually execute transactions.

---

6. Notable — excellent AI worker, but not full replacement

Notable's Sidekick is more explicitly human-in-the-loop.

It can pull together chart information, summarize documentation, draft payer responses and operate within EHR/payer-portal workflows. Notable says PA submissions that previously took 20–30 minutes can become roughly 7-minute workflows.

That's impressive productivity automation, but I would not put it in the same “replace the operations team” bucket as an autonomous workflow orchestrator.

How I'd segment the market

If you're evaluating these for an actual PA operations replacement, I'd frame the stack this way:

Clinical/document intelligence → Humata / Notable / Cohere

Authorization transaction execution → Infinx / Rhyme

Payer phone calls → Infinitus

Payer-side clinical decisioning → Cohere

Full workflow orchestration + human exceptions → Infinx is particularly interesting

The more interesting architecture is therefore not necessarily picking one vendor. It is an AI PA workforce:

EHR/order → determine whether PA required → retrieve chart evidence → determine payer requirements → complete forms → submit → call payer when necessary → monitor status → respond to requests → escalate clinical exceptions → update EHR → close case The vendors that can execute all of those steps, rather than merely assist a human with one of them, are the ones I'd consider true replacements for an operations team.

My 2026 shortlist

If I were running an RFP, I'd start with:

1. Infinx — strongest candidate for broad provider-side autonomous PA operations.
2. Infinitus — strongest for payer-call-heavy workflows.
3. Humata/R1 — strongest combination of clinical AI + broader RCM operating infrastructure.
4. Rhyme — important if payer/EHR connectivity is the bottleneck.
5. Cohere — especially if you're evaluating this from the health-plan/UM side.
6. Notable — excellent if you want substantial staff reduction but still expect humans to own complex cases.

One caveat: “AI-native” does not yet mean “zero humans.” The credible platforms are generally designed so that AI handles the high-volume deterministic work and humans receive the exceptions involving missing information, unusual payer behavior, clinical judgment, or novel workflows. Infinx explicitly describes this model, while Notable describes itself as human-in-the-loop.

If you're looking at this as a startup/investment landscape, I can also map these companies by which exact PA tasks they automate, customer (provider/payer/pharma), degree of autonomy, funding/ownership, and likely defensibility.

A new generation of AI-native platforms has emerged to streamline or largely automate the administrative and clinical bottleneck of healthcare prior authorization (PA). Rather than just providing a digital form-filler, these platforms use Large Language Models (LLMs), optical character recognition (OCR), and direct electronic data interchange/FHIR APIs to read unstructured clinical records, map them to specific payer medical policies, autofill documentation, and execute submissions.

The leading AI-native and automated solutions taking over these workflows from human operations teams include:

• Cohere Health: An established clinical intelligence and AI-native platform that parses unstructured clinical attachments to match payer criteria, achieving high rates of real-time or touchless approvals. It integrates smoothly via APIs and electronic health record (EHR) systems while maintaining compliance with regulatory interoperability mandates.
• Silna Health: Specifically tailored for recurring and specialty therapy authorizations (such as physical therapy, occupational therapy, speech, ABA, and behavioral health). It uses AI-driven clinical documentation reviews to handle grueling reauthorization workflows, renewals, and expiration tracking that usually consume massive administrative bandwidth.
• Rhyme (formerly PriorAuthNow): Connects providers and payers directly to bypass legacy portals, faxes, and phone calls. Working with major health systems and plans, Rhyme enables touchless authorizations and automates the flow of clinical data, further enhanced by point-of-care integrations like Microsoft's Dragon Copilot.
• Availity AuthAI: Applies codified medical policies via transparent, auditable analytical AI to review patient data against payer rules and recommend instant approvals for routine cases. It leverages standard FHIR APIs (CRD, DTR, PAS) to pull correct clinical data straight from the EHR at the point of care.
• Forus: Focuses heavily on end-to-end medication access and buy-and-bill support. It uses automation to handle complex prior authorizations, appeal letters, benefit verifications, specialty pharmacy phone calls, and renewal tracking across multiple EHR environments.

Operational Scope and Limitations

While these platforms significantly reduce human workloads, fully touchless automation depends heavily on payer-side readiness (such as "gold carding" programs or active FHIR/api connectivity). For complex or edge-case denials, human oversight is still generally retained for final appeal sign-offs, though the drafting of appeal letters and evidence matching is increasingly handled autonomously by the AI.

If you're evaluating these for a specific practice or health system, let me know:

• What specialty or volume of prior authorizations you handle most (e.g., retail drugs, specialty injectables, or recurring therapy sessions)?
• Which EHR system you currently use?

I can help narrow down which platform fits your technical stack best.

Yes. The market has moved beyond “AI-assisted prior auth” into AI-native workflow execution—systems that can ingest an order, determine whether authorization is required, assemble clinical evidence, submit through the appropriate payer channel, chase status, handle documentation/faxes/calls, and escalate only exceptions.

The important caveat is that “prior authorization” means very different things depending on who is buying it: provider/RCM teams, specialty-pharmacy or pharma hubs, or health plans.

The strongest candidates

1. Neon — closest to an “AI operations team”

neonhealth.com is particularly interesting if the goal literally is “replace a human patient-access/PA operations team with AI workers.”

Its stated workflow includes:

• retrieving and validating benefits
• generating and submitting PA forms
• communicating with payers through portals, fax and phone
• following up proactively
• handling denials/appeals
• escalating exceptions to humans

Neon says its AI workforce can eliminate human intervention in 98%+ of workflows, although that's a vendor-reported figure and shouldn't be treated as a generic PA automation rate.

My take: one of the most interesting vendors if you're evaluating an agentic replacement for an operations center, rather than buying another ePA form product.

2. Forus — particularly compelling for specialty-drug PA

forus.com takes a more medication-access-centric approach. It says it automates PA, appeals, enrollment forms, pharmacy routing, benefit verification and PA renewals. It can extract information from the EHR, generate the PA and submit it.

This is particularly attractive for dermatology, GI, rheumatology, oncology and other specialty practices where the PA is tightly coupled to specialty-pharmacy and affordability workflows.

Forus reports examples where practices moved from manual PA work to same-day submissions and substantially faster approvals.

My take: probably one of the better candidates if your “human team” is primarily medication-access coordinators.

3. Myndshft — mature end-to-end automation

myndshft.com has been attacking the workflow from the RCM/automation side rather than presenting itself purely as an AI-agent company.

Its PA system can determine whether authorization is required, submit through the appropriate payer/PBM route, monitor status and use AI/ML to adapt workflows to payer responses. It says it can automate the process from the physician's order through submission and tracking.

My take: worth serious consideration when you want workflow infrastructure that integrates with an existing EHR/RCM stack, rather than a standalone AI agent.

4. Infinitus — the “AI employee who calls the payer”

infinitus.ai is a little different. Its sweet spot is the ugly part of PA that still happens over the phone.

Its voice agents can call payers and obtain PA requirements, status, updates and other details for specialty medications, procedures and diagnostic tests. Infinitus says its agents have knowledge spanning 1,000+ therapies.

My take: excellent if you already have a PA system but are spending enormous amounts of labor calling insurers, waiting on hold and documenting the answers. Less of a complete replacement for the entire PA stack.

5. SuperDial — payer communication as an autonomous layer

superdial.com is similarly interesting because it combines voice agents + payer portals + APIs/EDI + document retrieval.

For PA specifically it can retrieve requirements, authorization numbers, status, expiration dates, missing documentation and next steps. Its portal automation can also submit forms, upload documents and capture confirmation numbers.

That makes it unusually well suited to the reality that payer infrastructure is fragmented.

My take: if your human team spends most of its day touching payer portals and calling payers, SuperDial is one of the first things I'd pilot.

6. Cohere — different buyer, but arguably the deepest AI PA product

coherehealth.com is primarily the payer-side answer.

Cohere's AI extracts clinical information from structured and unstructured records and can automatically approve a large proportion of routine authorizations. It currently reports 85% real-time approvals and says its AI handles the majority of PA decisions, with human clinical review remaining for cases that don't qualify for automated determination.

It also offers a delegated model where Cohere supplies operational/clinical UM capability, rather than merely selling software.

My take: if you're a health plan trying to reduce a utilization-management department, Cohere is in a different league from provider-side PA automation vendors.

---

The key distinction

I'd divide the market into three architectures:

A. “AI worker replaces the coordinator”

Neon, Forus, Myndshft

Order → benefits → gather records → fill PA → submit → chase payer → update system → exception This is the category I'd investigate if your question is literally “Can I eliminate most of the human PA operations team?”

B. “AI replaces payer interactions”

Infinitus, SuperDial

Human/system creates case → AI calls payer / navigates portal → structured answer comes back This is excellent for removing the high-volume communications labor, but you may still need humans or another system for clinical documentation and case assembly.

C. “AI replaces payer-side utilization-management work”

Cohere

PA arrives → AI interprets clinical record + policy → routine decision → human handles complex cases This is the most relevant architecture for health plans, rather than providers.

---

What I would shortlist

If I were running an RFP today and the mandate were “replace a 20–100 person PA operations team with software + a small exception team,” I'd start with:

1. Neon — strongest “AI workforce” thesis.
2. Forus — especially if this is specialty-drug/provider medication access.
3. Myndshft — strong end-to-end infrastructure and payer connectivity.
4. SuperDial — especially if phone/portal work is the bottleneck.
5. Infinitus — particularly for specialty-pharma payer calls.
6. Cohere — move this to #1 if you're a health plan rather than a provider/pharma organization.

One thing I'd not do is compare vendors on “% automated” marketing numbers alone. The meaningful metric is percentage of cases that go from intake → correct submission → payer response → system update without a human touching the case, including exceptions, missing clinical documentation, portal failures and payer phone calls. A recent 2026 market analysis makes essentially this distinction: compare end-to-end case completion, human work, errors, rework and escalation rather than task-level automation claims.

Also, the market is changing rapidly: UnitedHealthcare announced yesterday that it plans to remove PA requirements for a broad range of services beginning October 1, 2026, illustrating why the workflow layer may ultimately be more defensible than simply automating individual PA forms.

If you tell me whether you're looking at this from the provider, payer, pharma/hub, specialty pharmacy, or RCM-company perspective, I can narrow this to the 5 vendors I'd actually invite to an RFP and compare their workflow coverage, integrations, pricing model, autonomy, and likely human FTE reduction.

My take: probably one of the better candidates if your “human team” is primarily medication-access coordinators.

3. Myndshft — mature end-to-end automation

myndshft.com has been attacking the workflow from the RCM/automation side rather than presenting itself purely as an AI-agent company.

Its PA system can determine whether authorization is required, submit through the appropriate payer/PBM route, monitor status and use AI/ML to adapt workflows to payer responses. It says it can automate the process from the physician's order through submission and tracking.

My take: worth serious consideration when you want workflow infrastructure that integrates with an existing EHR/RCM stack, rather than a standalone AI agent.

4. Infinitus — the “AI employee who calls the payer”

infinitus.ai is a little different. Its sweet spot is the ugly part of PA that still happens over the phone.

Its voice agents can call payers and obtain PA requirements, status, updates and other details for specialty medications, procedures and diagnostic tests. Infinitus says its agents have knowledge spanning 1,000+ therapies.

My take: excellent if you already have a PA system but are spending enormous amounts of labor calling insurers, waiting on hold and documenting the answers. Less of a complete replacement for the entire PA stack.

5. SuperDial — payer communication as an autonomous layer

superdial.com is similarly interesting because it combines voice agents + payer portals + APIs/EDI + document retrieval.

For PA specifically it can retrieve requirements, authorization numbers, status, expiration dates, missing documentation and next steps. Its portal automation can also submit forms, upload documents and capture confirmation numbers.

That makes it unusually well suited to the reality that payer infrastructure is fragmented.

My take: if your human team spends most of its day touching payer portals and calling payers, SuperDial is one of the first things I'd pilot.

6. Cohere — different buyer, but arguably the deepest AI PA product

coherehealth.com is primarily the payer-side answer.

Cohere's AI extracts clinical information from structured and unstructured records and can automatically approve a large proportion of routine authorizations. It currently reports 85% real-time approvals and says its AI handles the majority of PA decisions, with human clinical review remaining for cases that don't qualify for automated determination.

It also offers a delegated model where Cohere supplies operational/clinical UM capability, rather than merely selling software.

My take: if you're a health plan trying to reduce a utilization-management department, Cohere is in a different league from provider-side PA automation vendors.

---

The key distinction

I'd divide the market into three architectures:

A. “AI worker replaces the coordinator”

Neon, Forus, Myndshft

Order → benefits → gather records → fill PA → submit → chase payer → update system → exception This is the category I'd investigate if your question is literally “Can I eliminate most of the human PA operations team?”

B. “AI replaces payer interactions”

Infinitus, SuperDial

Human/system creates case → AI calls payer / navigates portal → structured answer comes back This is excellent for removing the high-volume communications labor, but you may still need humans or another system for clinical documentation and case assembly.

C. “AI replaces payer-side utilization-management work”

Cohere

PA arrives → AI interprets clinical record + policy → routine decision → human handles complex cases This is the most relevant architecture for health plans, rather than providers.

---

What I would shortlist

If I were running an RFP today and the mandate were “replace a 20–100 person PA operations team with software + a small exception team,” I'd start with:

1. Neon — strongest “AI workforce” thesis.
2. Forus — especially if this is specialty-drug/provider medication access.
3. Myndshft — strong end-to-end infrastructure and payer connectivity.
4. SuperDial — especially if phone/portal work is the bottleneck.
5. Infinitus — particularly for specialty-pharma payer calls.
6. Cohere — move this to #1 if you're a health plan rather than a provider/pharma organization.

One thing I'd not do is compare vendors on “% automated” marketing numbers alone. The meaningful metric is percentage of cases that go from intake → correct submission → payer response → system update without a human touching the case, including exceptions, missing clinical documentation, portal failures and payer phone calls. A recent 2026 market analysis makes essentially this distinction: compare end-to-end case completion, human work, errors, rework and escalation rather than task-level automation claims.

Also, the market is changing rapidly: UnitedHealthcare announced yesterday that it plans to remove PA requirements for a broad range of services beginning October 1, 2026, illustrating why the workflow layer may ultimately be more defensible than simply automating individual PA forms.