Online blood labs return a table. Vitrubo reads it.

Most direct-to-consumer lab testing returns a table: marker, value, unit, reference range, a flag when the value falls outside it. A reading adds what the table leaves out. Which flags belong together. Whether a value has moved since the last draw. What medical guidance says about a result like this one. And which test is the best next step, if any. The table stays; the reading is placed beside it.

Yes. Direct-to-consumer testing rarely comes from one laboratory for long, and results from different laboratories arrive in different formats, units and panel names. Every value is mapped to one of 17,000+ LOINC codes and to UCUM units, so a lab test run by one online lab this year and another last year sit on one line. Trends read across labs, and a marker that looks normal in each single blood test can be seen drifting across them.

The laboratory's reviewing scientist or a clinician, as today. Vitrubo drafts the comment in the laboratory's style and cites the guidance behind it; the draft appears at the release step in the LIS or portal. The professional edits or approves. Vitrubo does not diagnose, does not select treatment and does not replace the person who releases the result.

Every statement links to named medical guidance: EHA, WHO, NICE, KDIGO, BSH and others, up to 30 sources behind one biomarker view. Numbers, units, ranges and flags are computed by deterministic code, never written by a model. Several AI models cross-check each other before a draft is shown. In a clinical program, roughly 100,000 anonymised hospital records were read on premises; on Stanford MedAgentBench the system completed 600 of 600 tasks across five runs.

Any lab test that returns a value: blood tests, urine tests and allergy tests arrive as values, from a full blood count and a lipid panel to a thyroid test, an iron study or a vitamin test. A laboratory test with a narrative result, a hospital record or a prescription is read as context rather than as a number. At-home testing kits posted back to a laboratory are read the same way as a draw at a site. The test menu of an online lab rarely contains a marker the mapping has not seen: 17,000+ LOINC codes cover what laboratory testing reports today.

PDF, photos and scans, HL7, FHIR R4 and JSON through the API. Blood, urine and allergy panels arrive as values. Hospital records, prescriptions and notes are read as context, so a lab test ordered online is read beside the medication list and the history the person chooses to add.

Under its own brand. White-label PDFs with the company's logo and disclaimers, a role-based portal for the laboratory, the clinician and the customer, or JSON through the API into the app the company already runs, with the report regenerated after each new test. Two views of the same analysis: plain language for the customer, full reasoning for the clinician. Vitrudoc, an AI chat grounded only in the person's own record, can sit beside the report.

The report regenerates with every new result. Flagged results are grouped into four zones: needs attention, worth watching, back to normal, always normal, with the next round of testing named where the guidance calls for one. Previous versions stay as history, so the person and the reviewing scientist can see what the reading said before the second draw and what changed after it. Repeat testing becomes a trend rather than two separate test reports.

A HIPAA business associate agreement for US deployments, a GDPR Article 28 processing agreement and Standard Contractual Clauses. Encryption in transit and at rest. Identity never enters the prompt: models see values and an internal ID. Deployment is in containers on your infrastructure or a private cloud, and you own the keys; Vitrubo ships versioned builds and has no access to the data.

Read it against the ranges printed on it, then bring it to a doctor. The live reports on this page show what a full reading looks like: flagged values grouped into stories, trends across draws, plain language with the guidance behind each statement. A medical test result is best read beside the history it belongs to, so bring earlier results too. For a person, Vitrubo is educational context to take to an appointment, not medical advice. It does not diagnose and does not tell you what to take.

The laboratory work is the same: a sample, an analyser, a reference range. Medical testing bought online and a medical test ordered by a doctor differ in who reads the result. A medical test ordered by a doctor comes back to someone who knows the history; a lab test ordered online often comes back to the person alone. The reading layer closes that gap on both sides: the laboratory releases a commented result, and the person receives a report written to be understood.

With a call, then sandbox access and API documentation. Your team sends its own anonymised panels in their real formats, your reviewing scientist validates the drafts, and nothing is committed until the output holds up on your own lab tests. Production follows when you are ready, inside your perimeter or a private cloud.