Documentation
Concepts and guides for building with Aqylon Inc. AI models, compute and communications APIs.
Developer preview. Endpoints, parameters and SDK methods shown here are illustrative and may change before general availability. Access is provided on request.
Getting started
Aqylon Inc. APIs are HTTPS and JSON interfaces. Every request is authenticated with an API key and every response includes a request ID you can quote to support.
To begin, request developer access. Once your organization is approved you can create keys, call the APIs from the sandbox and move to production when you're ready.
curl https://api.aqylon.co/v1/health \
-H "Authorization: Bearer $AQYLON_API_KEY"Authentication
Send your API key as a bearer token in the Authorization header. Keys are scoped to an organization and an environment (sandbox or production) and can be restricted to specific products.
Keep keys on your server. For browser or mobile clients, issue short-lived session tokens from your backend instead of embedding a key in the app.
Authorization: Bearer $AQYLON_API_KEYVoice: speech generation and recognition
Text-to-speech turns text into natural audio in a chosen voice and format. Speech-to-text transcribes recorded or streaming audio, with optional speaker diarization and translation.
Custom and cloned voices require documented consent from the voice owner and are reviewed before they can be used.
curl https://api.aqylon.co/v1/voice/tts \
-H "Authorization: Bearer $AQYLON_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "text": "Welcome to Aqylon Inc.", "voice": "sample-voice", "format": "mp3" }'Real-time conversations
Real-time sessions combine streaming speech recognition, a conversational model and speech generation over a single WebRTC or WebSocket connection, with interruption handling and voice activity detection.
Sessions can call your functions, hand off to a human agent and emit telemetry for each turn.
Video generation and understanding
Video jobs are asynchronous: submit a job, then poll it or receive a webhook when it completes. Generation supports text, image or video inputs; understanding returns scenes, captions and searchable descriptions.
Every model is labeled with its origin, first-party or partner, and its license terms.
curl https://api.aqylon.co/v1/video/jobs \
-H "Authorization: Bearer $AQYLON_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "task": "captions", "input_url": "https://example.com/clip.mp4" }'Models and inference
The Model Cloud exposes models through one API. List the models available to your organization, then call them by ID for serverless inference or provision a dedicated endpoint.
Rate limits and usage are reported per key, per model.
curl https://api.aqylon.co/v1/models -H "Authorization: Bearer $AQYLON_API_KEY"GPU compute
GPU resources are provisioned to your organization as instances, clusters or reserved capacity, with container deployment, Jupyter access and object storage.
Configurations, locations and pricing are agreed with our team before capacity is allocated.
Messaging, calls and phone numbers
Communications APIs cover in-app chat, WebRTC calling, SIP interconnect, messaging and one-time-passcode verification.
PSTN calling, phone numbers and SMS are available only in markets where carrier agreements and the required regulatory approvals are in place. Call recording requires consent from all parties where the law requires it.
curl https://api.aqylon.co/v1/messages \
-H "Authorization: Bearer $AQYLON_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "channel": "chat", "to": "user_123", "text": "Your order is ready." }'Webhooks
Register an HTTPS endpoint to receive events such as job completion, call status and message delivery. Each delivery is signed so you can verify it came from Aqylon Inc.; failed deliveries are retried with backoff.
POST https://your-app.example.com/aqylon-events
Aqylon-Signature: t=1730000000,v1=…
{ "type": "video.job.completed", "id": "evt_…", "data": { "job_id": "job_…" } }Errors and retries
Errors use standard HTTP status codes with a JSON body containing a machine-readable code, a message and the request ID. Retry 429 and 5xx responses with exponential backoff; use an Idempotency-Key header on POST requests so retries are safe.
{ "error": { "code": "rate_limited", "message": "Too many requests", "request_id": "req_…" } }Security and data handling
All traffic uses TLS. Keys can be rotated at any time and are never shown again after creation. Data you send is processed to provide the service; retention and deletion options are documented per product.