> ## Documentation Index
> Fetch the complete documentation index at: https://docs.olira.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# What is Olira?

> The patient context layer for clinical AI. Integrate once; your agents and workflows read complete, current patient state over API or MCP.

<Note>
  New to Olira? The [Quickstart](/quickstart) takes you from API key to
  compiled patient state in about ten minutes. If you are integrating an
  existing system, start with [Setup & data configuration](/setup).
</Note>

Olira is the patient context layer for clinical AI. Connect your backend and integrations once, and Olira ingests every health log (symptom reports, lab results, medication updates, wearable readings, check-ins) and continuously compiles them into one structured patient state. Your agents and workflows read that state over API or MCP, so every decision is made with the complete, current patient picture.

Every team building clinical AI hits the same wall: fragmented sources, hand-built pipelines, stale and incomplete context. Olira owns that substrate (normalization, temporal reconciliation, provenance) so your team never writes a data pipeline, and one integration serves every agent and workflow you ship. You control what each agent sees through views: templates you configure in the Console that shape patient state into exactly the context each agent and workflow needs. The more data flows in, the more capable your agents become.

## How it works

The core loop has four steps:

| Step | What you do                                | What Olira does                                                                                                            |
| ---- | ------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------- |
| 1    | Register a patient via the SDK or API      | Creates a patient record, returns an Olira-assigned ID (store it).                                                         |
| 2    | Submit health logs as they happen          | Ingests each log (whether you submitted it or it arrived via an integration) and updates the patient's state in real time. |
| 3    | Configure view templates in the Console    | Applies your templates to shape how state is presented to agents.                                                          |
| 4    | Connect agents or workflows via API or MCP | Delivers complete, current patient context on every query.                                                                 |

## Key concepts

Four terms appear throughout these docs. Here is what each one means.

| Concept             | What it is                                                                                                                                                                                                                                                              |
| ------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Patient**         | A record in Olira representing one person. Olira assigns the ID; you persist it in your database and use it for all future calls.                                                                                                                                       |
| **Log**             | A health data record submitted for a patient. Each log has a type (e.g. `symptom_report`, `medication_list_update`, `lab_results`) and a structured payload. Logs are the raw input; Olira compiles them into state.                                                    |
| **Patient State**   | The continuously-updated, structured representation of a patient, compiled from all their logs. Every new log triggers a state update in the form of an event. Your agents always read from the current state (never from raw logs).                                    |
| **View & template** | **Templates** are what you customize in the Console (plain language; output as Markdown, JSON, or FHIR): they specify how each view is constructed from patient state. A **view** is the timestamped result of applying a template (what agents and workflows receive). |

## Get started

<Columns cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
    Create an API key, send your first log, and read compiled patient state
    end to end.
  </Card>

  <Card title="Setup & data configuration" icon="sliders" href="/setup">
    Provision your organization, invite your team, and select the log types
    your platform will produce.
  </Card>

  <Card title="Send data" icon="arrow-up-from-bracket" href="/send-data/how-logs-work">
    How logs work, logging from your codebase, historical backfill, and
    integrations.
  </Card>

  <Card title="Read data" icon="database" href="/read-data/patient-state">
    Patient state ontology, views and templates, and accessing state from
    agents and workflows.
  </Card>
</Columns>

## Reference

<Columns cols={2}>
  <Card title="Log types catalog" icon="list" href="/reference/log-types/symptom-reports">
    Payload schemas, required fields, and source information for every
    supported log type.
  </Card>

  <Card title="Python SDK" icon="python" href="/reference/sdk">
    Method-by-method reference for patients, logs, ingestion, tokens, and
    state reads.
  </Card>

  <Card title="MCP Patient State" icon="plug" href="/mcp-server">
    JSON-RPC tools, resources, and prompts for agent runtimes.
  </Card>

  <Card title="CLI" icon="terminal" href="/cli">
    Authentication, API key management, JSONL validation, and ingestion jobs
    from the terminal.
  </Card>
</Columns>

<Info>
  HIPAA compliant · SOC 2 Type II audited · encrypted in transit and at rest
</Info>