Every healthcare provider knows the feeling. The patient visit ends, and then the real work begins: notes, prior auth requests, discharge paperwork, coding reviews, and coordination calls. The clinical moment lasted fifteen minutes. The documentation that follows takes three times as long. Multiply that across a health system, a physician group, or a skilled nursing facility, and the cumulative toll on capacity and clinician well-being becomes one of the defining operational challenges in U.S. healthcare today. Curee AI was built by a team that understood this problem from the inside. Founded in 2024 and based in San Francisco, Curee builds AI workflow automation tools designed exclusively for healthcare providers. The company's flagship product, Curee Workstation, acts as an intelligent layer running alongside a care team's existing EHR, handling the documentation-intensive, criteria-matching, coordination-heavy work that currently falls to clinical and administrative staff.
There is no shortage of technology being sold into healthcare. What is scarce is technology that actually changes how much time clinicians spend on non-clinical work. Most tools add a new interface, a new login, a new workflow, and the net effect is more administrative burden, not less. Curee is built on a different premise. The platform acts as a silent sidecar alongside the tools a care team already uses. It reads the same records, follows the same workflows, and surfaces completed work for a one-click review. Clinicians do not learn a new system. They simply stop doing certain kinds of mechanical work by hand. The automation is invisible; the time savings are not. The company is led by founder and CEO Karthik Rajakumar and backed by Google for Startups. Curee Workstation is HIPAA-compliant, SOC 2-certified, and built on zero-trust architecture, meeting the security and compliance standards healthcare enterprises require before deploying any AI tool in a clinical environment.
Curee Workstation is an agentic AI platform for acute and post-acute healthcare enterprises. It automates the workflows that consume the most time without adding clinical value: prior authorization submission, PDPM documentation review, admission packet parsing, discharge planning, and real-time chart scanning for documentation accuracy. The platform adapts to each organization's specific payer rules, clinical protocols, and operational structure. It ingests a facility's own guidelines, what Curee calls its walled garden, and produces outputs that reflect how that organization actually works, not a generic industry standard. A health system's prior auth workflow looks different from a skilled nursing facility's. Curee configures to both.
Curee does not require backend EHR integration. It works alongside Epic, Cerner, PointClickCare, and other major platforms through a native driver approach, no custom APIs, no IT project, no six-month implementation timeline. Organizations deploy in days and begin capturing value in the first billing cycle. This deployment model matters because the gap between signing a software contract and seeing results is where most healthcare technology investments stall. Curee closes that gap by design. Learn more about Curee Workstation.
What does Curee AI actually automate in a healthcare setting?
Curee automates the documentation-intensive workflows that consume clinical and administrative staff time without requiring clinical judgment: prior authorization drafting, PDPM chart scanning, admission packet parsing, discharge documentation, and real-time comorbidity identification. Staff review and approve the output; the AI handles the mechanical assembly.
How is Curee AI different from other healthcare AI tools?
Most healthcare AI tools add a new interface to an already complex workflow environment. Curee runs alongside existing systems as a silent sidecar, reading the same records staff already create, producing completed work for review, and requiring no new documentation habits or system logins.
What EHR systems does Curee AI work with?
Curee Workstation is compatible with Epic, Cerner, PointClickCare, and other major EHR platforms. It operates without backend integration, so compatibility does not depend on the EHR vendor's cooperation or a custom API project.
Who founded Curee AI?
Curee AI was founded by Karthik Rajakumar and launched in 2024. The company is headquartered in San Francisco and operates as a remote-first organization with a team spanning healthcare AI, clinical practice, and enterprise software engineering.
What security certifications does Curee AI hold?
Curee AI is HIPAA-compliant, SOC 2-certified, and built on a zero-trust security architecture. The platform meets the compliance requirements that healthcare enterprises expect before deploying AI in clinical or administrative workflows.
A prior authorization denial is rarely the end of the story. It is the beginning of a second, equally labor-intensive process: the appeal, the peer-to-peer, the resubmission, or the disposition change. Each of those follow-up steps consumes physician time, care coordinator hours, and, in the case of an inpatient stay, additional acute bed days while the authorization process plays out. The true cost of a denial is rarely what appears on the denial notice. Curee AI's prior authorization automation is designed to prevent denials at the front end rather than manage them on the back end. The system generates complete, criteria-matched prior authorization submissions from the patient's existing clinical record, stages them for one-click review, and submits them with the documentation payers need to approve on the first pass.
Healthcare administrators often measure prior authorization burden in minutes per request. The actual cost runs deeper. When an authorization is denied and a peer-to-peer review is required, an attending physician is pulled out of clinical time, sometimes for 30 to 60 minutes, to discuss a case with a payer medical reviewer who has access to less clinical context than the treating team. That is among the most expensive uses of physician time in a health system. Downstream, prior auth delays in the inpatient setting translate directly into extended length of stay. A patient who is clinically ready for discharge to a skilled nursing facility but is waiting for prior authorization remains in an acute bed that another patient needs. Health systems absorb the marginal cost of that day, and the patient's recovery environment is suboptimal. The authorization delay costs everyone.
Payer criteria matching in real time.
The most common reason prior authorizations are denied is not clinical; it is documentation. The patient meets the clinical criteria for the requested service, but the submission does not reflect that clearly enough for the payer reviewer to approve. Curee maps the patient's chart to payer-specific criteria language in real time, ensuring the submission is written the way payers expect to see it.
Always-current payer requirements
Payer criteria change frequently, and manual tracking of those changes is impractical at scale. Curee ingests updated payer guidelines continuously, so every submission reflects current criteria — not the requirements from the last time someone on the team checked.
Complete documentation packages
A prior auth denial for missing documentation is entirely preventable. Curee assembles the full supporting documentation package, clinical notes, diagnostic results, history, and functional assessments into the submission at the point of generation. When the submission arrives at the payer, everything they need to approve it is already there.
Review and approval by clinical staff
Curee does not submit prior authorizations without human review. Every package is staged for a clinical staff member to review and approve. The AI handles the assembly; the human handles the sign-off. This workflow keeps compliance intact and gives clinical staff a final check on every submission.
What causes most prior authorization denials?
The majority of prior authorization denials result from incomplete documentation, which does not match payer-specific criteria language or arrives without the supporting clinical evidence the payer needs to approve. Clinical criteria are often met — the submission simply does not demonstrate it clearly enough.
How does prior authorization automation reduce peer-to-peer reviews?
Peer-to-peer reviews are typically triggered when a payer reviewer cannot find sufficient clinical justification in the submitted documentation. When submissions are complete and criteria-matched from the start, reviewers have fewer grounds to request peer reviews. Facilities using Curee report meaningful reductions in physician time spent on insurance calls.
Can prior authorization automation handle Medicare Advantage plans?
Yes. Curee AI is configured to reflect each organization's payer mix, including Medicare Advantage plans with their varying criteria and documentation requirements. The system applies the relevant payer rules to each authorization based on the patient's coverage.
How long does it take to see results from prior authorization automation?
Because Curee deploys in days rather than months, most organizations see the impact within their first billing cycle. Fewer denials mean fewer peer-to-peer reviews and appeals, which frees staff time immediately.
Does prior auth automation work for both inpatient and outpatient settings?
Yes. Curee Workstation handles prior authorization workflows across health systems, physician groups, skilled nursing facilities, and inpatient rehab facilities, covering both inpatient level-of-care authorizations and outpatient procedure and service authorizations.
The transition to the Patient-Driven Payment Model was supposed to reward clinical complexity. Under PDPM, a skilled nursing facility that accurately documents a patient's diagnoses, functional status, and ancillary care needs receives payment that reflects the true cost of caring for that patient. In practice, many SNFs are not capturing the full picture — not because the clinical complexity is absent, but because the documentation process was not designed to find it systematically. Curee AI's PDPM automation addresses this gap by scanning 100% of patient charts in real time, surfacing missed documentation opportunities before the MDS window closes, and staging them for clinical review and approval. The result is reimbursement that accurately reflects patient acuity and revenue that was already being earned but not being captured.
PDPM calculates reimbursement across five components: physical therapy, occupational therapy, speech-language pathology, nursing, and non-therapy ancillaries. Each component is driven by specific clinical characteristics that must be documented in the MDS. When those characteristics are present in the patient's record but not reflected in the MDS coding, the facility receives a lower per diem than the patient's acuity supports. The documentation gaps that drive undercoding fall into three consistent patterns. The first is ICD-10 specificity: a diagnosis coded at a general level when a more specific code, reflecting laterality, etiology, or stage, would trigger a higher payment classification. The second is missed comorbidities: active conditions documented in physician notes or nursing assessments that do not make it into the MDS. The third is NTA underscoring: qualifying conditions for the non-therapy ancillary component that are present in the record but not flagged during MDS completion.
Systematic chart review, not spot-checks
Manual documentation review in a busy SNF is necessarily selective. A charge nurse or MDS coordinator reviewing charts at the end of a shift is looking at a subset of the record under time pressure. Curee reviews 100% of the chart, every physician note, nursing assessment, therapy evaluation, and lab result, and flags every item that has documentation implications for PDPM scoring.
Structured prompts for the interdisciplinary team
Rather than producing a report that someone has to translate into action, Curee surfaces structured prompts to the relevant member of the interdisciplinary team. A nursing documentation gap goes to the nursing team. A PT classification question goes to therapy. Each prompt identifies the specific finding in the chart, explains its PDPM implication, and requests the clinician's confirmation or correction.
MDS window awareness
Curee tracks each patient's MDS assessment window and prioritizes documentation prompts accordingly. Opportunities that are approaching a deadline surface first. The system does not flag issues after the window has closed; it identifies them early enough to act on them.
Compliant, defensible coding
Every documentation change prompted by Curee requires clinical confirmation. The system identifies what the chart supports; the clinician decides whether it accurately reflects the patient's condition. This workflow produces coding that is both accurate and defensible, grounded in the clinical record and confirmed by the treating team.
Skilled nursing facilities using Curee AI have collectively captured more than $20 million in previously undercoded revenue. On a per-facility basis, the revenue impact depends on census, payer mix, and the baseline accuracy of existing documentation processes — but the pattern is consistent: systematic AI review finds documentation opportunities that manual processes miss, and those opportunities translate to higher per diem reimbursement on the cases already in-house. The revenue is not new money. It is reimbursement; the facility was already earning through the care it was already delivering, care that was not being fully reflected in the documentation submitted for payment. PDPM automation does not change what care is delivered. It changes how completely the delivered care is documented.
What PDPM components does Curee AI help optimize?
Curee AI supports documentation accuracy across all five PDPM components, PT, OT, SLP, nursing, and NTA, with a particular focus on the nursing and NTA components, where ICD-10 specificity and comorbidity documentation have the greatest impact on per-diem rates.
How does Curee AI handle MDS documentation without replacing the MDS coordinator?
Curates surfaces documentation opportunities and structured prompts to the interdisciplinary team, but does not complete or submit MDS assessments. The MDS coordinator and clinical team review every prompt and make the coding decisions. Curee eliminates the chart review labor; the clinical expertise stays with the humans doing the work.
What is NTA scoring, and how does automation help?
The Non-Therapy Ancillary component of PDPM rewards documentation of specific high-cost clinical conditions. Curee identifies qualifying NTA conditions in the patient's chart that may not be captured in the current MDS coding and flags them for clinical review. Accurate NTA scoring can meaningfully increase per diem reimbursement on complex cases.
Is PDPM automation appropriate for smaller SNFs with limited IT resources?
Yes. Because Curee requires no backend EHR integration, there is no IT infrastructure requirement. The platform runs alongside PointClickCare or other SNF EHR systems without involving the facility's IT team in a deployment project. Small and mid-size facilities benefit as much as large operators.
How quickly does PDPM automation pay for itself?
Most SNF operators using Curee see the platform pay for itself within the first month of deployment, as documentation gaps are closed on active patient cases. The revenue impact shows up in the first billing cycle.
Hospital readmissions within 30 days are among the most visible quality and financial metrics in U.S. acute care. Under Medicare's Hospital Readmissions Reduction Program, health systems face payment reductions for excess readmissions in specific clinical categories. Beyond the penalty, an avoidable readmission represents a failure in care continuity: a patient who left the hospital without adequate support for what came next. The root cause of most avoidable readmissions is not clinical failure during the inpatient stay. It is a failure in the discharge planning process: the wrong post-acute placement, inadequate handoff documentation, or an unidentified risk factor that should have triggered additional follow-up before discharge. Curee AI's care transitions capability is built to address all three.
Discharge planning at scale is a documentation and coordination problem. A care coordinator managing 20 active patients cannot manually review every chart in depth, cross-reference every payer's authorization criteria, and build complete handoff packets for every pending discharge simultaneously. Something gets missed: a functional limitation that affects placement eligibility, a payer criterion that the current documentation does not satisfy, a high-risk condition that should trigger additional monitoring. The consequence is predictable: patients go to the wrong level of care, authorizations come back denied, and high-risk patients are readmitted within 30 days. Each of these outcomes is costly in a different way, to the health system's revenue, to the patient's recovery, and to the facility's quality metrics. And each is preventable with better information earlier in the discharge planning process.
Admission packet parsing at scale
A discharge that begins with incomplete information about the patient's history and needs is unlikely to end well. Curee reads full admission packets — which routinely run 15 to 150 pages — in under 60 seconds, producing a structured summary of nursing needs, active diagnoses, functional status, risk factors, and care requirements. Care coordinators start the discharge planning process with the complete picture, not the parts they had time to find manually.
Disposition prediction and approval probability
Curee analyzes the patient's clinical record against historical payer approval data to estimate the probability that a given post-acute disposition will receive authorization. When one disposition pathway is significantly more likely to be approved than another, the system surfaces that information before the placement decision is made — reducing the likelihood of a denial that delays discharge and extends the inpatient stay.
Readmission risk flagging before discharge
Curee identifies high-risk patients before they leave the building. The system analyzes clinical indicators, social risk factors embedded in the chart, and post-acute placement variables to flag patients who are statistically more likely to be readmitted. That flag goes to the care team while there is still time to act — by adjusting the discharge plan, adding a follow-up appointment, or escalating to a transition-of-care navigator.
Structured handoff documentation
The quality of care a patient receives after discharge depends partly on the quality of information the receiving facility receives with them. Curee generates structured handoff documentation that organizes the patient's nursing needs, medications, wound care protocols, fall precautions, and active diagnoses into a format the receiving team can act on immediately. The coordinator reviews and approves the package; the mechanical assembly is done.
For health systems operating near capacity, discharge planning efficiency is directly linked to throughput. Every day, a patient who is clinically ready for discharge remains in an acute bed because the post-acute authorization is pending, the handoff documentation is not finished, or the right placement has not been identified, is a day that the bed is unavailable to the next patient who needs it. Curee customers unlock an average of 3 to 5 days of capacity per case — not by discharging patients faster than clinically appropriate, but by removing the administrative delays that extend stays beyond clinical need.
How does AI-powered discharge planning reduce hospital readmissions?
Curee reduces readmissions by identifying high-risk patients before discharge and flagging the specific risk factors driving that risk. It also improves placement accuracy — ensuring patients go to the level of care their clinical needs require — and produces complete handoff documentation, so the receiving facility has the information needed to provide continuous, appropriate care.
What is the Hospital Readmissions Reduction Program, and how does discharge planning affect it?
The Hospital Readmissions Reduction Program reduces Medicare payments to hospitals with excess 30-day readmission rates for certain conditions, including heart failure, pneumonia, and hip and knee replacement. Improved discharge planning reduces avoidable readmissions by ensuring patients are placed appropriately, receive complete handoff documentation, and are flagged if they need additional post-discharge support.
How does Curee AI identify post-acute disposition options?
Curee analyzes the patient's clinical record, functional status, and payer coverage against historical authorization data to identify the disposition pathways most likely to be clinically appropriate and authorized. It surfaces the recommended placement with supporting documentation and an estimated approval probability.
Does Curee AI assist with both hospital-to-SNF and hospital-to-home transitions?
Yes. Curee supports the full range of post-acute disposition pathways, including skilled nursing, inpatient rehab, long-term acute care, home health, and home without services. The system identifies the appropriate level of care based on clinical need and payer criteria.
How does structured handoff documentation improve post-acute care quality?
When a receiving facility gets complete, organized handoff documentation, its clinical team can begin appropriate care immediately without reconstructing the patient's history from raw records. Medication reconciliation, wound care, fall precautions, and nursing needs are documented in a structured format that transfers clinical knowledge across the care transition.