10 Best AI Referral Intake Tools

Quick answer: The best AI referral intake tools in 2026 read inbound referrals — most still arriving by fax — understand them, extract the data, and carry the patient toward a booked appointment without staff retyping anything. Honey Health leads for practices that want the full inbound workflow run end to end by an AI agent inside their existing EHR. Tennr is the venture-backed leader in document AI for faxed referrals; Notable and Coral bring AI agents to intake at scale; Infinx adds AI document understanding plus services; ReferralMD layers AI onto a mature referral platform; and par8o, Updox, ReferWell, and Treatspace are established platforms adding intelligence to routing, communication, and scheduling. The right pick depends on how much of the work you want the AI to actually do.

Referral intake is one of the best targets for AI in the entire medical back office, because the work is high-volume, pattern-heavy, and trapped in documents. A referral arrives as a fax or a portal message, and a human has to read it, figure out what it's for, pull out the patient's demographics and insurance, decide whether anything's missing, and move the patient forward. That is precisely the kind of language-and-document task that modern AI — large language models and document-understanding models — has gotten genuinely good at, which is why a wave of well-funded companies has converged on exactly this problem.

But "AI referral intake" spans a real range, and the differences matter more than the shared label. Some tools use AI narrowly, to read a fax and extract fields, then hand the structured data back to staff. Others use AI agents to carry the whole intake forward — registering the patient, chasing missing records, verifying benefits, and scheduling — with people stepping in only on the exceptions. And some established referral platforms are adding AI features to workflows that were previously manual. Knowing where a tool sits on that spectrum is the difference between buying a smarter inbox and buying back staff hours.

This guide ranks the AI-native and AI-forward tools for inbound referral intake in 2026, with a clear best-fit and an honest read on where each one stops. It's the AI companion to our referral management software guide, and it's distinct from the outbound side covered in our AI referral submission tools guide.

Last updated: June 2026.

Where AI changes referral intake

AI touches referral intake at three distinct points, and a tool's value depends on how many of them it covers. The first is document understanding: reading a multi-page faxed referral — often a messy scan — and making sense of it. This is where the current generation of AI is strongest and where the category began, with models that can identify a referral, classify it, and read its contents the way a person would rather than relying on rigid templates.

The second is data extraction and structuring: pulling the patient's name, date of birth, insurance, referring provider, and reason for referral out of that document and into structured fields, ready to drop into an EHR. The third, and the hardest, is autonomous workflow execution: actually doing the downstream work — registering the patient, requesting missing records, verifying benefits, and scheduling — rather than just preparing it for a human. The most capable tools reach all three and add a fourth quality on top: knowing when to stop. A high-dollar or ambiguous referral should route to a person, and an intake AI that escalates the genuinely hard cases rather than forcing them through is far safer than one that automates everything blindly.

How we evaluated these AI referral intake tools

We included tools built around modern AI — large language models, machine learning, document-understanding models, or autonomous agents — that do real inbound referral intake work, serve US healthcare in 2026, and are HIPAA-compliant. We weighted these dimensions:

• AI depth — is the product AI-native, or an established platform adding AI features?
• How far the AI carries the work — read and extract only, or register, chase, verify, and schedule?
• Document and channel handling — can it read faxed referral packets, not just clean portal data?
• Exception handling — does it route the genuinely hard cases to a person?
• EHR fit and deployment effort — does it work with your systems without a long integration?

There's no universal winner. A fax-swamped specialty practice and a large health system automating patient access need different tools, so each entry carries a clear best-fit and an honest note on its limits.

AI referral intake tools at a glance

The 10 best AI referral intake tools in 2026

1. Honey Health — best for end-to-end AI inbound referral processing

Honey Health is built around a simple but demanding idea: an AI staff member that doesn't just read a referral but works it, the way a human intake coordinator would, end to end. The company makes trained, dedicated AI workers that log into a practice's existing systems and run administrative workflows on their own, and inbound referral intake is one of its core jobs. What sets the underlying technology apart is its agentic browser automation — not rules-based RPA, not an API integration, not a browser extension. Each AI worker runs in a virtual browser, logs in with its own credentials, reads and understands the full screen, and operates the application directly, so it handles dynamic screens, popups, and interface changes that break scripted bots, rewriting its own approach when an app changes. That reliability traces to a founding team that built anti-bot and automation systems at LinkedIn and Microsoft, where behaving like a real human user at scale was the entire job.

On referral intake, that translates into the full inbound flow rather than a slice of it. The agent identifies the incoming referral — frequently handed straight off from Honey's own fax-triage AI, which is how many referrals first arrive — checks for duplicates and whether the patient is new or returning, sets the patient up in the EHR, extracts and uploads the documentation, runs the authorization flow, faxes the referring provider to chase any missing records, verifies benefits, records the referral in the practice's CRM or SharePoint, coordinates templated scheduling and patient SMS, and follows up, qualifying each referral against the practice's checklist along the way. It does this inside the 20-plus EHRs a practice may already use, plus payer portals and any fax inbox, with no integration project. Honey reports 99.8 to 99.9 percent task accuracy, a HIPAA-compliant and SOC 2 platform, go-live in two to three weeks with no onboarding fees, and a "needs human review" queue for low-confidence cases, backed by a dedicated human success and technical team that tunes the workflow and ships changes, often same day.

Scheduling is where the workflow stops short of fully autonomous: Honey handles templated scheduling and patient SMS, but fully dynamic, back-and-forth appointment negotiation isn't there yet, so a practice whose intake depends on intricate scheduling will keep a person on that step. Pricing is per task and works out to roughly three to six dollars per hour of equivalent human work, with customers citing 2.91x savings per dollar and 80 to 95 percent less manual effort. Among AI referral tools, the distinction is reach: many read and extract, but Honey is built to carry the whole intake and escalate only the exceptions. For practices that want inbound referrals genuinely worked rather than merely digitized, inside the systems they already run, it's the most complete option here.

2. Tennr — best for AI intake from faxed referrals

Tennr is the company most associated with bringing modern AI to referral intake, and its story explains why. Founders who met at Stanford doing AI and large-language-model research built the platform on a deliberately contrarian premise — that the way to fix healthcare referrals wasn't to replace the fax machine but to "power" it with AI. Investors agreed emphatically: Tennr raised $18 million in April 2024 and then a $101 million Series C in June 2025, backed by Lightspeed and Foundation Capital, one of the larger recent bets on healthcare back-office AI.

The product is a patient-orchestration platform whose document AI, trained on medical records and processing more than 10 million documents a month, reads inbound referrals arriving by fax, email, or e-portal, extracts and structures the information, runs eligibility and benefits, and moves a patient from "fax to first visit" faster. Tennr reports compressing pre-visit processing from weeks to hours and cutting front-end billing errors by 98 percent at customers, and in March 2026 it added an automated voice-AI calling feature to handle the outbound phone work intake generates. Its focus is squarely the receiving provider, with three stated goals: more completed referrals, fewer preventable denials, and a more efficient team.

Tennr's document-understanding depth is genuinely strong, and a practice whose pain is the faxed-referral pipeline will find it tailored to exactly that. The considerations are that it's a substantial platform — possibly more than a small practice wanting a lighter tool needs — and that its fast, funding-fueled expansion into payment posting, claims auditing, and records management means a broad and still-evolving surface. Best for referral-heavy practices that want best-in-class AI reading and processing inbound faxed referrals at volume.

3. Notable Health — best for AI intake automation at scale

Notable Health is among the most heavily funded AI automation companies in healthcare, built around a platform of AI agents that automate patient access, revenue cycle, and care operations. Based in San Mateo, it raised a $100 million Series B led by ICONIQ Growth in November 2021 — roughly $123 million in total, with Greylock, F-Prime, Oak HC/FT, and Maverick also backing it — and has concentrated on the high-volume administrative work that large provider organizations face: registration, intake, scheduling, authorizations, and patient follow-up.

For referral intake, Notable's strength is the breadth and scale of its intake automation. Its AI agents absorb the registration, data capture, and outreach that referrals create, and because the same platform spans patient access and revenue cycle, a large organization can apply it across many adjacent workflows rather than buying a single-purpose referral tool. For a health system standardizing on one automation layer, that consolidation is the draw.

The orientation toward enterprise scale is also the boundary. Notable is built for larger health systems and provider groups with the volume to justify a platform deployment, so a small specialty practice may find it heavier than a focused referral tool, and referral intake is one application of a wide platform rather than its sole specialty. Best for large health systems automating patient intake at scale, with referrals as one of many workflows.

4. Coral AI — best for AI that replaces intake RPA

Coral AI makes a specific argument: that the rules-based RPA many practices bolted onto intake over the past decade is the wrong tool, because it breaks the moment a document or screen deviates from its script. Coral replaces that brittle automation with AI agents that reason over documents and workflows, and referral and intake processing sits alongside its fax and document automation. It raised seed funding led by Lightspeed and reports running more than 500,000 workflows a month across its customers, a sign of real production volume for a young company.

On referral intake, Coral's AI extracts the clinical and demographic data from inbound documents, reasons about what's needed, and drafts or routes the next step, with a design that keeps a human in the loop where judgment is required. For an organization that has lived through the frustration of RPA that constantly breaks, the appeal is an automation layer that bends instead of snapping when a referral doesn't match the expected format.

As a seed-stage company, Coral's footprint is still building, it isn't referral-specific, and its output is frequently drafted for human review rather than fired fully autonomously on every case — a reasonable posture, but one that keeps people more involved than a fully autonomous agent. Best for practices replacing fragile intake RPA with adaptable AI across documents and referrals.

5. Infinx — best for AI document understanding plus services

Infinx pairs AI software with a human services team across patient access and revenue cycle, and referral intake fits its model through AI-powered document understanding for healthcare referrals — extracting, analyzing, and processing the medical data inside a referral — combined with augmented-intelligence virtual intake management. The blend of AI and staffed services is the company's defining characteristic: software handles what it can, and the services team handles the rest, so the customer sees completed work rather than just a tool.

For a practice or group that would rather offload referral intake as a managed capability than operate software, Infinx's services-plus-AI model is a genuine fit, and its established presence in patient access and prior authorization means the referral piece connects to adjacent workflows like eligibility and authorization that intake naturally touches.

The trade-off is that the services-led model is less a self-serve AI agent than a managed partnership, so the value scales with how much work you hand over, and a practice wanting to keep the workflow fully in-house may prefer a pure-software option. Best for groups that want AI document understanding for referrals delivered as a managed service alongside patient access.

6. ReferralMD — best for AI inside a mature referral platform

ReferralMD is a long-established referral management platform that has layered AI onto a deep, full-featured base rather than starting from AI alone. The Charleston-based company positions itself as an all-in-one AI solution for patient intake and referrals, connecting primary care, specialists, and health systems, and its AI shows up most concretely in AI faxing and intake automation — reading and routing inbound faxed referrals — within a platform that also covers closed-loop tracking, eConsults, scheduling, and analytics. It carries strong user-review scores and offers a free tier for inbound and outbound management plus a provider CRM.

For an organization that wants AI intake but also wants the surrounding referral-management machinery — stage tracking, leakage analytics, loop closure — in one mature system, ReferralMD's combination is compelling, since the AI accelerates intake while the platform manages everything around it.

Because the AI is an enhancement to an established platform rather than the product's foundation, its autonomy on the hands-on intake steps is more measured than a purpose-built agent's, and a buyer focused purely on cutting-edge document AI may find the AI layer less deep than a dedicated AI-native tool. Best for practices that want AI-assisted intake inside a comprehensive, proven referral platform.

7. par8o — best for intelligent referral routing

par8o brings intelligence to a specific slice of intake: routing a referred patient to the right in-network provider based on clinical fit, location, and availability. It has a long care-coordination track record — urgent-care operator CityMD adopted it to pair routing with care coordination, and hospital networks have used it for value-based initiatives — and in February 2026 it was acquired by NuvemRx from R1 RCM, sharpening its focus on helping covered entities capture specialty referrals and retain patients within their networks.

For intake, par8o's contribution is the matching-and-routing decision — getting a referred patient directed to the right place rather than leaking out — which is a meaningful piece of the intake puzzle for networks worried about retention, and an area where algorithmic matching genuinely outperforms manual judgment.

Its intelligence is concentrated in routing rather than the document-heavy front of intake, so reading a faxed referral and extracting its data still leans on other tools or staff, and the recent acquisition means the roadmap is being repositioned. It's better understood as smart routing than full AI intake. Best for networks that want intelligent referral routing and patient retention as part of intake.

8. Updox — best for AI-assisted intake in a unified inbox

Updox, part of EverHealth since EverCommerce acquired it in December 2020, brings referrals into the same inbox as a practice's fax, secure text, telehealth, and document management, and it has been adding automation and intelligence to that communication core. Its referral workflow lets a practice manage the entire referral process from start to finish inside the application, and because so many referrals arrive by fax, its HIPAA-compliant faxing-and-documents foundation puts inbound referrals where staff already work.

For a smaller primary care or specialty practice that wants AI-assisted referral handling without standing up a separate platform, Updox's all-in-one inbox is the appeal — referrals are managed alongside everything else, with growing automation to speed the routing and handling.

The honest framing is that Updox is a communication platform adding intelligence rather than an AI-native intake engine, so its automation is lighter than a purpose-built document-AI tool, and referral management is one capability among many. A referral-heavy specialty group needing deep autonomous intake will likely outgrow it. Best for independent practices that want AI-assisted referral handling inside a unified communication inbox.

9. ReferWell — best for automated intake plus scheduling

ReferWell automates the part of intake that most often determines whether a referral converts: the scheduling. Its ReferWell Connect platform orchestrates every step from a care recommendation to a booked appointment, pairing smart provider matching against in-network rosters with real-time booking and closed-loop tracking, and it serves both provider organizations and health plans, embedding into care-coordination workflows.

For an organization whose intake breaks at the scheduling gap — referrals received but never booked — ReferWell's automated matching and real-time booking attack exactly that failure point, and its payer-side footprint makes it relevant where keeping patients in-network has financial stakes.

Because its intelligence is concentrated on provider matching and scheduling, the document-heavy front of intake — reading a faxed packet and extracting its data — is less its strength, so a fax-swamped practice may pair it with a document-AI tool, and its dual provider-and-payer orientation can mean more involved implementations. Best for organizations whose intake bottleneck is automated matching and scheduling.

10. Treatspace — best for automated closed-loop intake

Treatspace brings automation to the discipline of the closed loop, building its referral management around connecting providers so patients reliably get appointments and consult reports return to the referring physician — its stated aim of closing the loop "100% of the time." It's an award-winning platform pairing high-performance referral management with patient engagement and provider-network tools, with a following among independent practices and networks formalizing their referral relationships.

For intake, Treatspace's value is automated tracking and loop closure — ensuring each inbound referral is followed through to completion and reported back — which is precisely the step that quietly breaks and erodes referral relationships, and automating it protects both patients and referral sources.

As a focused referral platform adding automation rather than an AI-native document engine, Treatspace handles tracking and loop closure well but leaves more of the document-heavy intake to staff, and it's a smaller player than the funded AI entrants. Best for practices that want automated, dependable closed-loop intake with their referral partners.

How to choose an AI referral intake tool

Start by deciding how far you want the AI to carry the work, because that single choice sorts this list. If you want AI to read a faxed referral and hand your team clean, structured data, document-AI tools and AI-enabled platforms will serve you. If you want the AI to actually do the intake — register the patient, chase the missing records, verify benefits, and schedule — you need an autonomous agent like Honey Health's that runs the whole flow and escalates only the exceptions. The first buys accuracy and speed at the data-entry step; the second buys back staff hours across the entire workflow.

Then weigh AI depth against platform maturity. The AI-native entrants (Honey, Tennr, Notable, Coral) lead on document understanding and autonomy; the established platforms adding AI (ReferralMD, Updox, par8o, ReferWell, Treatspace) bring proven referral machinery — tracking, analytics, loop closure, scheduling — with intelligence layered on. Neither is automatically better; the right answer depends on whether your gap is the raw intake labor or the surrounding management, and many organizations end up valuing one far more than the other.

Pressure-test how the tool handles documents and exceptions, since those are where AI intake earns or loses its keep. Confirm it can read the messy, multi-page faxed referrals you actually receive — not just clean portal data — because a tool that stumbles on faxes won't help a fax-heavy specialty. And ask what happens to the ambiguous or high-stakes referral: a tool that escalates the genuinely hard cases to a person is safer and more trustworthy than one that forces everything through automatically.

Finally, account for deployment effort. AI agents that operate your existing EHR and fax inbox, like Honey's, avoid the per-EHR integration tax that enterprise platforms carry, which for a practice is the difference between fixing this quarter's intake backlog and waiting on an IT roadmap. For the full field including non-AI platforms, see our referral management software guide, and for sending referrals out, the AI referral submission tools companion.

Frequently asked questions

What is AI referral intake?

AI referral intake uses machine learning, document-understanding models, or autonomous agents to handle inbound referrals — reading the referral (often a fax), extracting patient and insurance data, and moving the patient toward a booked appointment. Tools range from those that read and extract for staff to autonomous agents like Honey Health that register the patient, chase records, verify benefits, and schedule, escalating only the exceptions.

Can AI read faxed referrals?

Yes — this is where AI referral intake began and where it's strongest. Modern document-understanding models read multi-page faxed referral packets, including messy scans, and extract the data the way a person would rather than relying on rigid templates. Tennr built its platform around AI for faxed referrals; Honey Health pairs fax triage with referral intake so a fax flows straight into the workflow; and several platforms now offer AI faxing.

How much of referral intake can AI actually do?

It depends on the tool. Document-AI tools read and extract, then hand structured data to staff. Autonomous agents go further — registering the patient, requesting missing records, verifying benefits, recording the referral, and coordinating scheduling. Honey Health, for example, runs the full inbound workflow and routes only low-confidence cases to a person, while many tools stop at the data-extraction step.

Is AI referral intake accurate and safe enough to trust?

The safety comes from design. The better tools report high task accuracy and, crucially, keep a human in the loop on low-confidence or high-stakes referrals rather than forcing everything through. Honey Health reports 99.8 to 99.9 percent task accuracy and routes uncertain cases to a "needs human review" queue. When evaluating any tool, ask both about accuracy and about how it handles the cases it isn't sure about.

How is AI referral intake different from AI referral submission?

Intake is inbound: receiving and processing referrals that come to your practice. Submission is outbound: sending referrals out to specialists and following them to a booked appointment. The AI techniques overlap, but the workflows and buyers differ — a specialist's intake problem and a primary-care practice's submission problem are not the same. Our AI referral submission guide covers the outbound side.

How much do AI referral intake tools cost?

Pricing models vary. Agent platforms like Honey Health charge per completed task (netting to roughly three to six dollars per hour of equivalent work), so cost scales with volume; AI document and intake platforms price by subscription or per-document; services-led models like Infinx price by engagement; and established platforms adding AI price by subscription, sometimes with a free tier. Compare every option against the loaded cost of the staff time inbound referrals consume today.

AI has made inbound referral intake one of the most automatable workflows in the back office, but the tools differ sharply in how far they carry the work — from reading a fax to running the entire intake. Decide how much you want the AI to actually do, weigh AI depth against platform maturity, and insist on faxed-document handling and sensible exception routing. For a practice that wants inbound referrals worked end to end by an AI agent inside the systems it already runs, Honey Health is the most complete place to begin.