Getting Started

Start with the conceptual model on Workforce Decision Layer, then follow the integration flow below.

The public API base URL is:

https://api.avelinlabs.com

1. Create an Account

Use the Platform API for account and contract management:

POST /api/v1/platform/register
GET /api/v1/platform/verify-email?token=...
POST /api/v1/platform/login
POST /api/v1/platform/api-keys/create

Create a Runtime API Key for product endpoints and keep it secure.

2. Generate Runtime API Access

Use the management token from login to create runtime keys.

The free beta contract allows at most two active, unexpired Runtime API Keys.

3. Call Your First API

Start with these Runtime endpoints for workforce decision support:

POST /api/v1/occupation/candidates for a non-authoritative O*NET 30.3 candidate shortlist. It defaults to five candidates, never selects a mapping, and is separate from job classification.
POST /api/v1/job/classify for a single top occupation output.
POST /api/v1/job/analyze for ranked candidates, skill evidence, and review-relevant output fields.

Use runtime bearer token headers for product endpoints and keep credentials out of browser/client-side builds. See the occupation candidate examples for an executable request.

4. Add Customer Grounding (Optional)

Customer Grounding is included under current free-tier contracts when available.

A practical sequence is:

check grounding capabilities,
register source,
ingest text/Markdown or supported file types,
generate grounded role intelligence,
inspect report confidence, uncertainty, and traces.

5. Integrate and route by policy

Build application behavior around decision-support outputs:

evaluate confidence and uncertainty,
preserve request identifiers,
route low-confidence outcomes to human review,
keep unsupported outputs out of irreversible automation.

6. MCP integration

The current MCP model is a customer-operated stdio adapter over authenticated Runtime API capabilities.

Configure MCP with runtime API key and base URL, then treat returned outputs as governed decision-support signals.