> For the complete documentation index, see [llms.txt](https://docs.bloodgpt.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.bloodgpt.com/getting-started/readme.md).

# Overview

AI-powered blood test analysis API for healthcare professionals and laboratories

BloodGPT is a custom multi-agent system powered by three specialized LLMs, designed to interpret lab results with medical-grade accuracy. It turns complex lab data into clear, actionable health insights through a simple API.

## Key Features

* **Instant Analysis** — Process blood tests across 98 languages and different lab formats
* **Flexible Input** — Accept HL7, PDF, images, or structured JSON data
* **White-Label Reports** — Generate reports with your own branding, logo, and signature
* **Scalable API** — Built for high-volume laboratory and clinic workflows
* **Real-time Webhooks** — Receive instant notifications when processing completes
* **Batch Processing** — Submit up to 100 tests in a single API call

## Use Cases

* **Labs** — Automated lab result analysis at scale
* **Clinics** — Streamlined patient result interpretation
* **Healthcare Platforms** — Integrate blood analysis into your existing systems

## How It Works

1. **Submit** a blood test — PDF, image, HL7, or structured JSON — via the [Processing API](/submitting-tests/process.md).
2. **Process** — BloodGPT recognizes and interprets the test asynchronously.
3. **Retrieve** results — poll the pre-signed `download_url`, or receive a [webhook](/interpretation-and-results/webhooks.md).
4. **Deliver** to patients — as a white-label PDF, a hosted [Patient Portal](/integration-guides/integration-options/patient-portal.md), or embedded [Widgets](/reference/widgets.md).

## Get Started

{% hint style="info" %}
**New here?** Start with the [Quick Start](/getting-started/playbook.md) — it helps you request access and pick the integration that fits you (PDF-only, patient portal, widgets, or full API).
{% endhint %}

**Ready to build?** Head to the [Processing API](/submitting-tests/process.md) to submit your first test, then explore:

* [Upload](/submitting-tests/upload.md) — submit documents and extract structured data (JSON / FHIR)
* [HL7 Format](/submitting-tests/hl7.md) — submit HL7 v2.x laboratory messages
* [Batch Processing](/submitting-tests/batch.md) — submit up to 100 tests at once
* [Patients API](/patients-and-catalog/patients.md) · [Webhooks](/interpretation-and-results/webhooks.md) · [Widgets](/reference/widgets.md)
* [Errors & Validation](/reference/errors.md) — error codes, validation rules, and accepted date formats

## Authentication

All requests require your API key as a Bearer token:

```
Authorization: Bearer YOUR_API_KEY
```

To obtain an API key, [request access](https://platform.bloodgpt.com/request-access) or contact us at <contact@bloodgpt.com>.

## Privacy & Security

* **EU processing region** — Production workloads, databases, object storage, and clinical records run in Google Cloud `europe-west4` (Netherlands). Cloud SQL, object storage, and caches are encrypted at rest with customer-managed keys (CMEK); all API and portal traffic is HTTPS/TLS only.
* **Clinical records in a dedicated FHIR store** — Patient observations, reports, and interpretations are held in per-organization Google Cloud Healthcare API FHIR stores. Tenant isolation is enforced across API authentication, storage prefixes, and Healthcare datasets.
* **No training on your data** — We do not use customer or patient data to train or fine-tune models. Interpretation runs on third-party foundation models (OpenAI, Google Vertex AI / Gemini) called through our own gateway, which is configured not to persist prompt or response bodies. Application-level AI tracing goes to a HIPAA-scoped environment under an executed BAA.
* **Subprocessors, not open sharing** — We do not sell your data or use it for advertising. Delivering the service does involve a defined set of subprocessors, each receiving only the data class its function requires — the identity provider gets authentication data and no health data, the payment provider billing metadata and no health data, the workflow orchestrator internal identifiers and status and no clinical payload. The current list is published at [platform.bloodgpt.com/legal/subprocessors](https://platform.bloodgpt.com/legal/subprocessors), and which of them apply to you follows from the features you enable.
* **Contracts** — We act as a processor for our customers under a GDPR Data Processing Agreement (DPA), and sign Business Associate Agreements (BAAs) with subprocessors that handle PHI where HIPAA applies. HIPAA has no certification scheme, so ask us for the current DPA, BAA position, and subprocessor schedule for the integration you are planning.

{% hint style="info" %}
**Running a vendor or security review?** Contact <contact@bloodgpt.com> for the DPA, subprocessor schedule, and a data-flow description scoped to the features you intend to enable.
{% endhint %}