Toy examples
Hello-world demos rarely survive contact with real traffic.
Guides
From first call to production confidence
Practical, opinionated guides for the patterns that matter: structured output, routing strategies, safe retries, batch processing and spend control.
Proven patterns
for common workloads
Real request shapes
using the actual API
Production ready
retries, limits and spend
The problem
What is slowing you down today
Scattered advice
Best practices live in threads, posts and old tickets.
Surprises at scale
Retries, limits and costs only show up when volume does.
Capabilities
Everything you need, nothing you don't
Structured output
Ask for JSON with named fields, set temperature to 0, and validate before you store.
Routing strategies
When to send a capability, when to pin a model_id, and when to set an account default.
Safe retries
Retry 429, 502 and 503 with backoff and the same Idempotency-Key.
Batch processing
Pace workers under your rate limit and key each item for idempotent metering.
Prompt hygiene
Stable prompts get cache hits. Keep variable data at the end and boilerplate consistent.
Spend control
Cap max_tokens, use lower-priced models for routine work and watch the usage dashboard.
Use cases
What teams run on it
01
Document extraction
From raw text to validated JSON.
02
Support assist
Draft replies grounded in your own policy text.
03
Agent workflows
A model per step, with fail-over and cost per step.
04
Content at scale
Batch generation with predictable cost.
How it works
From sign-up to production
- 1
Pick a guide
Start with the workload closest to yours.
- 2
Try it in the playground
Run the prompts before writing code.
- 3
Ship it
Copy the request shape into your integration.
Control
Built so you stay in charge
Keys you control
API keys are shown once at creation and stored only as a hash. Revoke any key instantly from Settings.
Spend you can see
Every request carries its own receipt: tokens in, tokens out, latency and price. Nothing is billed that you cannot trace.
Limits you set
Per-account rate limits and a free monthly allowance keep experiments safe before they become production.
Data you decide on
You choose exactly what you send. Repeat prompts can be served from your cache, and cache hits are never billed.
For developers
One endpoint. Every model.
Send a prompt and a capability, or name a model. One bearer key, one small JSON body, and a full receipt on every response.
curl https://YOUR-DOMAIN/api/ainf/inference \
-H "Authorization: Bearer $AI_INFERENCE_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: order-1042" \
-d '{"prompt":"Summarise this ticket in one line","capability":"chat"}'Questions, answered
Do the guides use the real API?
Yes. Every example uses POST /api/ainf/inference with the real body fields and headers.
Are the guides language-specific?
No. The API is plain HTTPS and JSON, so the patterns apply to any language.
Where do I start?
With the quickstart, then the guide closest to your workload.
Can I request a guide?
Yes. Email our team with the workload you want covered.
Learn the patterns, ship the feature
Start with the quickstart and build from there.