Integrations
Model-agnostic, by construction.
AetherLab operates at the data layer: send text or images and receive policy verdicts. Use scalar checks in live paths or server-side batches for bulk moderation and backfills, regardless of which model produced the content.
One check, three places to put it.
Inbound
Before the model
Screen user prompts before they reach your model: injection attempts, prohibited requests, abuse patterns.
Outbound
After the model
Screen generated text and images before they reach users: policy violations, unsafe claims, brand breaks.
Anywhere
Any content flow
The checks accept any data, AI-generated or human, so moderation queues, UGC, and agent pipelines use the same policy line. Existing corpora can use asynchronous batches for review or backfills.
Python-first. REST underneath. Scalar or batch.
Use aetherlab 0.5.0 on PyPI or call REST from any language. Scalar checks use /v1/guardrails/prompt and /v1/guardrails/media; recommended /v1/guardrails/prompt/batches and /v1/guardrails/media/batches routes accept simple items plus shared settings, with /v1/batches retained for advanced generic and JSONL workflows.
Guardrails are one pip install aetherlab away. Use check_prompt and check_media for scalar checks, or post simple items plus shared settings to the guardrail-specific batch routes for server-side bulk moderation and backfills.
curl -sS -X POST \
https://api.aetherlab.co/v1/guardrails/prompt/batches \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"items": [
"historical text to screen",
"a second prompt to review"
],
"settings": {
"blacklisted_keyword": "guaranteed returns",
"reasoning_mode": "medium"
}
}'Bring your stack. Keep your policies.
If your content flows over HTTP or already sits in a review corpus, it can be checked. Tell us about your architecture and we'll map the scalar and batch integration points.
Ask about the Evidence Pack
Leave your email and we'll walk you through what an Evidence Pack contains for your use case: severity-scored findings, business-impact mapping, and the approval record.