Skip to main content
This guide assumes your organization, event types, and API keys are already configured (Console and/or CLI; see Setup). Here we wire your codebase to the Olira Python SDK: initialize the client, register one or many patients, and emit logs that drive patient state.

​
Install tooling

CLI (optional: keys, login, MCP config): 
brew install olira-ai/tap/olira
# or: curl -fsSL https://install.olira.ai | sh
Python SDK (required for ingestion from your backend): 
pip install olira
# or: uv add olira

​
Authenticate in your service

Use an API key with at least sdk:event-log and api:manage-patients for creating patients and logging. Create the key in the Console or with olira keys create (see the CLI key commands). Store it in a secret manager and pass it at runtime: 
import os
import olira

olira.init(api_key=os.environ["OLIRA_API_KEY"])
# or set OLIRA_API_KEY and call olira.init()

​
Initialize the SDK and create patients

​
Single patient

patient = olira.create_patient(
    first_name="Ada",
    last_name="Example",
    email="ada@example.com",
    external_identifiers=[{"system": "mrn", "value": "MRN-12345"}],
)
patient_id = patient.id

​
Multiple patients

Use create_patients_batch to register up to 500 patients in one POST /v1/patients/batch call. Pass a list of CreatePatientRequest objects; the response is a PatientBatchResult with items (successes, each with index and id) and errors (failures per index); partial success is supported. 
from olira import CreatePatientRequest, ExternalIdentifier

result = olira.create_patients_batch([
    CreatePatientRequest(
        first_name=row["first_name"],
        last_name=row["last_name"],
        email=row.get("email"),
        external_identifiers=[ExternalIdentifier(system="mrn", value=row["mrn"])],
    )
    for row in enrollment_rows
])
for item in result.items:
    # item.index → position in input list, item.id → Olira patient id
    ...
for err in result.errors:
    # err.index, err.code, err.message: handle or retry failed rows
    ...
Build the request list from a comprehension, a CSV import, or any iterator. For a single registration you can still use create_patient. Anchor rules and field shapes match CreatePatientRequest in the Python SDK reference.

​
Log data

Use olira.log() or olira.log_batch() to send structured logs: an OliraLogType and a payload that matches each log type (see the log types catalog). Use the correct patient_id for each log, whether you run one patient or many. 
from olira import OliraLogType

olira.log(
    log_type=OliraLogType.SYMPTOM_REPORT,
    patient_id=patient_id,
    payload={"instrument": "esas_r", "symptoms": [{"name": "pain", "score": 3}]},
)
olira.flush()
Optional OliraTrace links a log to an object in your system (conversation id, questionnaire id, etc.) for provenance in observability and when reading state later.

​
Verify ingestion

When your organization has observability enabled, use Observability → Logs and Events in the Console, and Patient detail to confirm modules and views update as expected. See Console observability.