Healthcare Administrative Automation
P

Predoc

Predoc is a New York company whose AI agents run medical record operations for care providers and clinical research sites, from retrieving outside records to filing inbound faxes and turning both into usable clinical data. Five agents divide the work. A Research Agent picks the retrieval route from a proprietary provider registry, querying health information exchanges, pharmacy and imaging networks and EHR connections, and a Voice AI Agent trained on more than three years of transcribed calls phones and faxes the offices that are not connected. An Indexation Agent splits and categorizes incoming documents, an Extraction Agent pulls clinical data from PDFs, scans and handwritten notes, and a Curation Agent normalizes, maps and deduplicates it into a longitudinal patient record.

Predoc Curated Data, launched in August 2026, delivers that record into an EHR, a data warehouse or a web app, and a REST API with FHIR output, webhooks and a clinical summary endpoint serves software teams. Customers include Oshi Health, Science 37, New York Cancer and Blood Specialists, Nebraska Cancer Specialists, Revival Research and Doctronic. Oshi saves about 30 minutes per request, and Science 37 cut time to consent by 50%. Predoc claims a 95%+ completion rate with turnaround from instant to 48 hours.

Founders include chief executive Nishant Hari, chief technology officer Alex Daniels and chief medical officer Dr. Kaushal Kulkarni. Memr, Inc. operates the company, which raised $30 million in 2025 in a round led by Base10 Partners.

AI Health Index verifiedNovember 19, 2026
Compare Predoc with other vendors
Founded
2022
Headquarters
New York, New York
Website
www.predoc.ai/
Categories
healthcare-admin-automation, clinical-summarization, clinical-trials-ai
Indexed Products
Predoc Web App, Predoc API, Record Retrieval, Fax Ingestion, Predoc Curated Data, Clinical Summary API
Buyer Segments
Healthcare providers, Health information management teams, Clinical research sites, Healthcare software teams
Assessment

Capability Axes

An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read

AI Capability
AA on AI CentralityThe artificial intelligence is the product. Remove the model and there is nothing left to sell.
Vendor Published

Agents do the work Predoc sells. The Research Agent picks the retrieval route, the Voice AI Agent calls provider offices, the Indexation Agent splits and files documents, the Extraction Agent reads PDFs, scans and handwriting, and the Curation Agent turns the result into a deduplicated longitudinal record. The web app and API both sit on top of that engine. A customer can still request records and view a case in an interface, but without the agents there is no retrieval at scale, no structured output and nothing left to sell.

DD on Autonomy and Oversight ModelNo oversight structure is published. An absolute claim that the system does not err grades here too, because a buyer who believes it will not build the review step that would catch a failure.
Vendor Published

Agents query digital networks, chase records by phone and fax, index what arrives and deliver output straight into the EHR or a warehouse, while customers can also open organized documents and summaries in the web app, and webhooks flag when records are ready for review. The platform page promises that Predoc gathers "complete, longitudinal histories from any U.S. source" and the complete picture beyond what exchanges carry. That is a completeness claim.

Reviewing the retrieved set cannot reveal a record that was never found, nothing requires a person to check for gaps before direct delivery, and no routing threshold is described, even as the same page reports a 95%+ completion rate.

BB on Model and Technology TransparencyThe approach or the suppliers are named without the version and update discipline behind them. Naming a supplier is the entry to this band both here and on Model Supply Chain Disclosure, which ask different questions of the same disclosure: who receives the data, and what produces the output.
Vendor Published

The Voice AI Agent is trained on more than three years of transcribed retrieval calls, the Indexation and Extraction Agents on documents annotated by a team of human experts, and the Curation Agent on a model Predoc says clinicians and customers verify. A Research Agent chooses the retrieval route from a proprietary provider registry, and the platform combines OCR, clinical parsing and mapping to standard clinical vocabularies. That names the approach and the training data at a high level. The production model families, any outside providers, version numbers and how customers learn of a model change are not stated, and the annotation set is not characterized by source or document type.

CC on Model Supply Chain DisclosureThe architecture is described and no model provider is named. Naming a hosting provider alone does not lift a record out of this band. Record the host in the note, because it matters for residency and breach scope, and grade on the model layer, which is the question this axis is named for.
Vendor Published

Each agent handles one stage, from the Voice AI Agent's calls to the Extraction and Curation Agents' processing, and Predoc says the curation model is verified by clinicians and customers. Record sources are named by category: health information exchanges, pharmacy, imaging and EHR networks, and provider offices reached by fax and phone, with output delivered by API or into an EHR. The privacy policy allows disclosure to providers under appropriate agreements.

Which outside companies supply speech recognition, voice synthesis or the language models behind extraction, whether they receive identifiable records, and what retention and training terms they operate under are not stated.

BB on Clinical and Operational EvidenceNamed deployments with dated outcome figures and enough method to test them, or published research short of independent validation.
Vendor Published

Oshi Health used Predoc to manage record requests and split incoming gastrointestinal records into consults, procedures, imaging and labs, and reports saving about 30 minutes per request. Science 37 piloted retrieval in 2023 and reports a 50% cut in time to consent with fewer staff chasing records. Those are named deployments with dated outcomes and a described workflow.

Predoc's other figures, 75%+ faster turnaround across customers, a 95%+ completion rate and $2 million in annual savings at an unnamed customer, carry no method. No result isolates extraction accuracy or the share of records missed or misfiled.

CC on AI Safety and PHI StewardshipGeneral assurances of privacy and security that do not answer the questions artificial intelligence raises: what is retained, what reaches a model, and what happens to it there.
Vendor Published

The service privacy policy allows information, including participant protected health information where applicable, to be processed for developing, training and validating AI models, subject to law and safeguards, and sets retention by purpose and legal obligation without a fixed window. The security program encrypts data and keeps patient identity separate from clinical data. Those are protective controls set beside an explicit training permission.

Nothing specifies deidentification before training, a retention period for inference data or a customer contract term that overrides the policy, and no process covers telling a client when a model output caused a safety problem.

Regulatory and Compliance
BB on HIPAA and BAA PostureBusiness associate status is stated and supported by a substantive privacy document, with the agreement or its scope not fully published. For a vendor outside the United States, an equivalent regime documented to this depth grades here. A vendor whose product does not process protected health information grades here too when it says so plainly and explains the scope, with a service privacy document behind it: stating a position a buyer can rely on is the posture this axis grades, and the band above is closed to it because there is no agreement to publish.
Vendor Published

Predoc handles identifiable records, including diagnoses, treatment details, medications and record identifiers, received from authorized sources over protected connections, and its privacy policy addresses those clinical records and the automated processing performed on them. The security program states HIPAA compliance and offers to sign a business associate agreement within 48 hours. That establishes business associate status with a service privacy document behind it. The agreement's text is not published, so how it applies to the web app, API and retrieval service and what it requires of each subcontractor are settled when it is signed.

BB on Security Certifications and Trust CenterA recognized certification is named in the vendor own material without the artifact, or with a scope or renewal question the buyer has to raise. A recognized certification here is an audit of how the vendor operates security, such as SOC 2, HITRUST or ISO/IEC 27001. A certification has a scope and a clock, and both are part of this grade. A certification the vendor attaches to its data center, hosting facility or cloud provider, rather than presenting as its own, is the host's and does not count here.
Vendor Published

Predoc holds a SOC 2 Type II report with annual independent audits covering security, availability and confidentiality, and qualified prospects can request it under a confidentiality agreement. The security program describes encryption, separation of customers, separation of patient identity from clinical data and US based servers and storage, and research customers get Part 11 electronic record and signature controls with a log of access and changes.

The homepage also carries CHIME Foundation and CMS Health Tech Ecosystem early adopter badges, which are recognitions rather than audits. The report sits behind a request, with no self service trust portal.

CC on FDA and Regulatory StatusNo device claim is made and the product is scoped accordingly. Most administrative and operational products sit here and are not penalized for it, because this axis grades the appropriateness of the positioning rather than possession of a clearance. A stated limit on use, such as research use only or a clause that the product gives no medical advice, beside marketing that sells the product for clinical decisions also sits here, with both quoted, because the buyer has to ask which one governs.
Vendor Published

Predoc retrieves and organizes records for visit preparation, care transitions, population management, medication reconciliation and research screening, normalizing notes and labs into a patient record and producing configurable clinical summaries. Its research customers get audit trails built for FDA 21 CFR Part 11, an electronic records requirement rather than a device position.

The product is sold as record operations and data infrastructure, organizing documentation that already exists, and makes no device claim, which fits that scope for retrieval and filing. The summaries and research screening support decisions by care and enrollment teams without being marketed as making them.

DD on AI Governance and Bias DisclosureNothing published on how model behavior is governed or tested. Multilingual operation with no subgroup performance sits here when the vendor markets recognition quality as a strength, because a caller the system failed to understand leaves no complaint and no record.
Vendor Published

Retrieval behavior comes from years of past calls to provider offices, extraction from expert annotated records, and users get a traceable view of normalized data. None of that shows how accuracy varies by record type, care setting, handwriting, language or patient group, or whether records for some patients are found less often than for others.

Predoc offers no comparison of misclassification or missed records across groups and no material a customer committee could use to evaluate those differences. The case studies describe time saved and staffing changes, which say nothing about where the agents fail.

CC on AI Liability and RecourseMechanisms exist that let someone challenge an output, such as audit trails, source traceability or review before commit, with nothing standing behind the output and no route for the harmed party.
Vendor Published

Processed records link back to their source documents, and the platform keeps an audit trail of access and changes built to FDA 21 CFR Part 11 for research customers, so a user reviewing a chart or summary in the web app can check a suspect extract against the original. The privacy policy gives individuals a route to request access to or amendment of their information, which covers a person's rights rather than the customer's. The terms of service page on the site is a placeholder, and no remedy is offered for records that were never retrieved, a document filed to the wrong patient or a wrong clinical summary.

Integration and Deployment
BB on EHR and Interoperability DepthNamed systems with read access or one directional writing, or standards support with named deployments behind it.
Vendor Published

A RESTful API takes retrieval requests and record uploads and returns FHIR structured data for medications, labs, imaging, procedures and provider notes, along with documents and clinical summaries, while webhooks signal when records are retrieved, processed and ready for review. Records arrive from health information exchanges, pharmacy and imaging networks and EHR connections, and go out to a customer's EHR or warehouse. Integration takes one to two developers two to four weeks. Oshi Health and Nebraska Cancer Specialists run retrieval and fax ingestion in their workflows, but no named EHR is shown receiving Predoc data in production at a named site.

BB on Deployment Model and Data ResidencyOptions and residency are stated with isolation or the processing path left open.
Vendor Published

The web app and the production API run on the same records engine, and Predoc states that its servers and storage are in the United States, with patient identity kept logically separate from clinical data and accounts separated by customer. Output can go to the customer's EHR or data warehouse, as an EHR export or as a PDF download. Delivery routes are not storage. Where any outside speech or model supplier processes records during retrieval and extraction is not stated, so the US commitment covers Predoc's own infrastructure and leaves the inference path open.

Commercial
CC on Commercial TransparencyNo price is published and the posture is discoverable: a buyer can establish how the product is sold and what drives the cost before contacting the vendor. Most of the index sits here.
Vendor Published

Customers start with a pilot on the web app and move to the production API on the same records engine, with integration estimated at one to two developers over two to four weeks. Buyers can request patient records, review documents and choose structured data, summaries, an EHR export or PDFs, and fax ingestion and curated data are offered as distinct workflows. The cost drivers are visible from that menu. The unit of charge, pilot cost, minimum retrieval volume and any fees for conversion or summaries are not published, and the commercial terms of service page is a placeholder.

BB on Setting and Specialty CoverageCoverage is named with validation behind part of it.
Vendor Published

Care providers and clinical research sites are the buyers. Oncology practices are prominent, including New York Cancer and Blood Specialists and Nebraska Cancer Specialists, which in October 2026 handed its inbound faxes and document management to Predoc's agents with outside retrieval to follow, and Oshi Health uses it for virtual gastrointestinal care.

Science 37, Revival Research and Clinical Enrollment use it for study screening and enrollment, and Doctronic and other software teams use structured records in their own products. The case studies validate retrieval and document handling in oncology, gastroenterology and research, while performance by condition, specialty and source organization is not reported.

Alternatives

Alternatives to Predoc

Other products that do the same kind of work as Predoc, and what each one does.

Metriport pulls a patient's records from health information exchanges, record systems, pharmacies, labs and admission and discharge feeds, then matches, deduplicates and normalizes them into one longitudinal record.

Abstractive Health retrieves a patient's records from the national exchange networks and summarizes them, with every sentence of the summary linked to the source note it came from.

DigitalOwl turns medical records into summaries, structured data and case level conclusions for the people who decide coverage, claims and litigation: life insurance underwriters and auditors, property and casualty claims teams, and law firms.

Valerie Health runs an AI front office for independent physician groups, combining software agents with its own operations staff.

Semble combines a practice management system and electronic health record for private healthcare organizations. Staff use it to manage bookings, tasks, reminders, invoices and patient records across single clinics and larger groups.

Patients.app, from Granville Labs, finds a patient's records across other hospitals and practices and uses AI to screen them for a specific decision.

Commercial

Pricing

Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.

Entry Price Pricing Basis BAA Tier Implementation Source
Custom quote
Predoc sells its record workflow through a pilot on the web app and a commercial agreement, with the production API, fax ingestion and curated data delivered on the same records engine. Predoc states HIPAA compliance and offers to sign a business associate agreement within 48 hours. Its scope across the web app, API and retrieval service is set at signing. Production API integration takes an estimated one to two developers over two to four weeks. No setup or professional services fee is listed. Vendor Published

No pilot price, unit of charge or minimum retrieval volume appears, and terms for retrieval, document processing and summaries are set commercially.