Planner model
Predicts, per provider, whether API, portal, phone or fax will succeed fastest, and in what order to try them.
Retrievant is a retrieval intelligence company. Where an API exists, we query it. Where one doesn't, our agents pick up the phone, log in to the portal, or read the fax. Everything comes back as one verified record, with the source on every line.
Reaching the systems where records actually live
Product
Every retrieval is planned by a model that decides, per provider, the fastest path to the record, then dispatches the right agent. Click through to see each one work.
Patient: Marie Bouchard DOB: 02/09/1984
Date of visit: 03/14/2019. Pt presents 6 wks post L4-L5 discectomy. Reports improvement in radicular pain.
Allergies: Penicillin (rash). Current meds: Atorvastatin 20 mg daily, ibuprofen PRN.
Plan: PT 2x/wk, f/u 6 wks. Consider MRI if sx persist
Watch
A retrieval, start to finish, at 20× speed.
00:00
00:12
00:24
00:41
00:58
Webhook fired to Northgate Legal · 38 min elapsed · every field linked to its source
Intelligence
Most retrieval breaks slowly: a portal redesign here, a changed fax number there. Retrievant treats every attempt as training signal. The context layer remembers how each of 90,000 providers prefers to be asked, so success rates rise the longer we run.
See our coverage in your regionPredicts, per provider, whether API, portal, phone or fax will succeed fastest, and in what order to try them.
Turns scanned charts, PDFs and free-text notes into structured FHIR resources, with confidence scores on each extraction.
Merges duplicates, resolves conflicting dates and doses, and flags what it couldn't settle for a human reviewer.
A living profile of every care site: hours, preferred channel, quirks, typical turnaround. Updated after every contact.
Use cases
Complete chronologies for demand letters and depositions, without paralegals on hold.
Learn more →Replace attending-physician statements with structured history in under an hour.
Learn more →Longitudinal, consented records for pre-screening and real-world cohorts.
Learn more →See a new patient's full story, including the surgery they forgot to mention, before the first visit.
Learn more →Give users real clinical context in-app with a single consent screen.
Learn more →Patients moving between the EU and US carry their history with them. We reconcile the formats.
Learn more →Trust
Autonomous doesn't mean unsupervised. Every agent action is scoped by the patient's consent, logged in full, and escalated to a human the moment it leaves the script.
Team
Retrievant began in Paris after one too many patients arrived with a shoebox of paper. The founding team has put AI in front of tens of millions of patients across Europe and the US.
Co-founder & CEO
Led product for the AI features inside France's largest appointment and records platform, used by most French GPs. Previously scaled a digital health insurer from launch to a million members. Trained as a clinical data engineer.
LinkedIn →Co-founder & CTO
Built hospital interoperability infrastructure connecting 700+ facilities, then led ML engineering at a biotech training models on real-world oncology data. Has written more HL7 parsers than he would like to admit.
LinkedIn →Questions
No. APIs cover the convenient third of a patient's history. The rest lives in places without one. Retrievant's agents are purpose-built for those channels, with a planning model that picks the route, task-specific voice and browser agents that execute it, and a clinical reader that structures the result. The model is the coordinator, not the whole product.
Yes, on every call, in the first sentence. It also identifies the patient, the requesting party, and the consent on file. Transcripts are stored with the retrieval.
One plain-language consent screen listing who is asking, what they'll receive and for how long. Afterwards, a dashboard of every retrieval made on their behalf, with one-click revoke.
Most teams make a first retrieval the same afternoon using the REST API and hosted consent flow. A full rollout with webhooks and branding takes about a week.
We tell you, precisely. The gap report lists where we looked, how, and why nothing came back, so you're never guessing whether the file is complete.
Pick a real case with a messy history. We'll run the agents live on a 30-minute call and show you what came back, and from where.
No slide deck. Just the record.