# Predoc launches Curated Data to turn faxed charts into AI-ready records > Source: > Published: 2026-08-27 17:56:34+00:00 # Predoc launches Curated Data to turn faxed charts into AI-ready records **The New York startup is extending its retrieval service into a normalized patient-data layer delivered through an app or API.** By [RuntimeWire Staff](/author/runtimewire-staff) ยท Published Primary source: [PR Newswire](https://www.prnewswire.com/news-releases/predoc-launches-curated-data-layer-to-make-fragmented-medical-records-usable-302861520.html) ## Why it matters Predoc is moving from chasing charts to owning the data layer beneath clinical AI, a larger and stickier role if its curation proves accurate at scale. [Predoc](https://www.predoc.ai/company/about?ref=runtimewire) launched [Curated Data](https://www.predoc.ai/platform?ref=runtimewire) on August 27, giving healthcare organizations a managed layer for turning records from exchanges, EHRs, pharmacies, imaging networks, scans and faxes into structured patient histories. The launch pushes Predoc deeper into the data infrastructure underneath clinical AI, where access to records matters far less if the information arrives duplicated, inconsistently labeled or buried inside a long PDF. Founder and CEO [Nishant Hari](https://thehealthcaretechnologyreport.podbean.com/e/automating-patient-record-retrieval-predoc-s-nishant-hari/?ref=runtimewire) came to that problem after nearly a decade in finance and derivatives trading at Citi and HSBC. His wife, a practicing physician, was spending time chasing records by phone and fax while patients arrived with binders of paperwork. Hari founded Predoc in 2022 with physicians Dr. Kaushal Kulkarni and Dr. Priya Mehta and technical co-founder Alex Daniels. Kulkarni encountered the same failure from inside the clinic. The neuro-ophthalmologist initially used Predoc to deal with large faxed charts, then invested in Predoc and joined as a co-founder, [according to Fortune](https://fortune.com/2025/09/02/predoc-raises-30-million-to-stop-document-chasing-in-healthcare/?ref=runtimewire). His progression from customer to backer to chief medical officer helps explain the shape of Curated Data: Predoc is building around the work required before a clinician can make use of a record, rather than treating document delivery as the finish line. In its [August 27 announcement](https://www.prnewswire.com/news-releases/predoc-launches-curated-data-layer-to-make-fragmented-medical-records-usable-302861520.html?ref=runtimewire), Predoc said Curated Data is generally available. Predoc says it has processed over 10 million pages of medical records. Its website also claims more than 1 million patients served, though Predoc has not published independent measures of extraction accuracy, reconciliation quality or the clinical effect of the new product. ### Healthcare has plenty of exchanged records and too little usable data The technical plumbing for moving healthcare records has expanded rapidly. As of February 2026, [ASTP/ONC reported that nearly 500 million health records had been exchanged through the Trusted Exchange Framework and Common Agreement](https://healthit.gov/news/tefca-americas-national-interoperability-network-reaches-nearly-500-million-health-records-exchanged-as-hhs-leverages-technology-and-ai-to-lower-costs-and-reduce-burden/?ref=runtimewire). That volume has not fixed the clinician experience. A [JAMA Network Open study](https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2841326?ref=runtimewire) found that fewer than 15% of physicians reported an ideal experience obtaining, finding and reconciling external data inside their EHRs. Predoc is betting that the remaining problem belongs in a layer between record exchange and the software that consumes the results. Curated Data extracts information from structured feeds and unstructured documents, maps clinical concepts to standard vocabularies, removes duplicates and reconciles conflicting entries. Predoc then organizes those facts into a longitudinal patient record containing items such as medications, laboratory results, imaging, procedures, encounters and provider notes. Customers can use Predoc's web application or connect through its REST API. Predoc says the API supports webhooks, document-level endpoints and FHIR-compatible output, allowing healthcare organizations to return the processed information to an EHR, analytics environment or clinical application. The harder part sits behind that interface. Predoc retrieves data from digital networks while also handling provider outreach, incoming faxes, PDFs, scans and handwritten material. Predoc's [platform materials](https://www.predoc.ai/platform?ref=runtimewire) describe human-in-the-loop validation as part of the service. That makes Curated Data a managed healthcare operation supported by software, rather than a simple API that returns whatever an exchange already holds. For Hari, that operational layer is part of the product. Predoc is asking customers to outsource record retrieval, document processing and data maintenance together. The approach creates a broader contract and more responsibility: Predoc has to gather missing material, interpret multiple formats, resolve duplicate clinical concepts and preserve links to the original records. ### The AI pitch starts before the model Predoc is positioning Curated Data as infrastructure for clinical AI without claiming that a language model can repair a fragmented patient history on its own. [Predoc's materials](https://www.predoc.ai/platform?ref=runtimewire) argue that summaries and question-answering tools do not create a normalized, auditable dataset that can be reused across a patient population. That distinction is commercially useful. Healthcare organizations experimenting with clinical copilots, patient monitoring, trial recruitment or risk analytics still need consistent inputs. A medication listed under a brand name in one record and a generic name in another has to be recognized as the same clinical concept. Measurements need comparable units. Duplicated encounters need consolidation. Each extracted fact also needs traceability to its source if a clinician or auditor questions it. [Predoc's internal analysis](https://www.predoc.ai/blog/beyond-access-performance-as-the-new-standard-for-hie-data?ref=runtimewire) of a July 2026 HIE dataset illustrates the amount of cleanup involved, though it should be read as Predoc's own measurement. Predoc said its pipeline retained 41.9% of roughly 60.7 million raw rows after normalization, semantic mapping, deduplication, consolidation and removal of records it considered non-informative. The number shows how much repetition can accumulate in exchanged healthcare data. It does not establish whether every removed or reconciled entry was handled correctly. Predoc is entering an established category. [Health Gorilla](https://developer.healthgorilla.com/docs/patient-chart?ref=runtimewire), [Zus Health](https://zushealth.com/platform/?ref=runtimewire) and [Particle Health](https://docs.particlehealth.com/docs/patient-data-apis?ref=runtimewire) also sell infrastructure for retrieving, standardizing or delivering longitudinal clinical data. Predoc's wager is that combining nationwide digital sources with direct retrieval from disconnected providers, fax ingestion and ongoing curation will produce a more complete record than digital exchange alone. ### A funded expansion from retrieval into infrastructure Predoc raised a combined $30 million across seed and Series A financing reported in September 2025. Base10 Partners led the financing, with participation from Northzone, Eniac Ventures, Entrepreneurs Roundtable Accelerator and other investors. Predoc's valuation and the split between the two rounds were not reported. At the time of that financing, Hari told Fortune that Predoc had about 35 customers and had generated roughly $500,000 in revenue during an earlier six-month period. Those figures described Predoc before the Curated Data launch and do not establish its current revenue or customer count. The financing gave Hari and his co-founders room to move Predoc beyond the original chart-chasing problem. Retrieval brought Predoc into the workflow; curation can make Predoc part of the customer's persistent data architecture. That is a larger opportunity, with a correspondingly higher burden of proof. Medical records can contain contradictory diagnoses, outdated medication lists and clinically meaningful duplicates. Cleaning them for software consumption requires judgment about which differences matter. Hari's core thesis remains grounded in a visible healthcare failure: clinicians cannot use information they cannot find, compare or trust. Curated Data turns that observation into a product category Predoc can sell across care delivery, research, analytics and AI. Predoc now has to show that its combination of automation and human review can maintain the accuracy, speed and provenance those uses demand.