Inconsistent APIs
Provider-specific conventions make changes ripple through product code.
THE CONTROL PLANE FOR AI APPLICATIONS
Route AI requests, monitor usage, and control costs across providers — through one developer-friendly gateway.
THE OPERATIONAL GAP
Every new provider adds another interface, another bill, and another place where reliability can slip.
Provider-specific conventions make changes ripple through product code.
Request volume, latency, and errors can be scattered across different tools.
Usage can grow before a team has a clear view of its costs and limits.
Retries, provider failover, and access management take time away from the product.
THE ROUTIQEL CONTROL PLANE
One project dashboard for the request path, provider settings, and usage signals. Available features are marked clearly.
Send OpenAI-compatible chat requests through one authenticated endpoint. Start with a deterministic sandbox, then configure a supported provider connection.
Learn more Local MVPChoose a primary provider, optional demo fallback, request timeout, and bounded retry count. Rules are stored per project.
Learn more Local MVPReview request status, provider, model, latency, token counts, trace ID, and estimated cost. Prompts are not stored by default.
Learn more EstimatedSee usage and estimated cost from the request events recorded by your local project. Provider prices can vary and estimates are not invoices.
Learn more Local MVPSet a monthly project limit and an alert threshold. Requests that exceed an enabled hard limit are rejected before forwarding.
Learn more OptionalConnect supported OpenAI-compatible provider accounts. Credentials are encrypted at rest by the local API and never returned to the browser.
Learn moreDEVELOPER EXPERIENCE
Keep your application code pointed at one endpoint. Start with the deterministic demo provider, then configure an authorized provider in your workspace.
import json
from urllib.request import Request, urlopen
payload = {
"model": "routiqel-demo",
"messages": [{"role": "user", "content": "Hello"}],
}
request = Request(
"http://localhost:8000/v1/chat/completions",
data=json.dumps(payload).encode(),
headers={"Authorization": "Bearer rtq_YOUR_PROJECT_KEY", "Content-Type": "application/json"},
method="POST",
)
with urlopen(request) as response:
print(json.load(response)["choices"][0]["message"]["content"])Requires an API key created in your Routiqel workspace. The demo adapter needs no provider credentials.
HOW IT WORKS
Connect an authorized provider, choose a route, send requests, and use the resulting metadata to understand your stack.
Configure an authorized provider, or use the local sandbox.
Set a primary provider and optional demo fallback.
Review request status, latency, tokens, and estimates.
Set a project budget and enforce a hard monthly limit.
Routiqel authenticates the project, applies limits, selects a route, and records request metadata. Local demo architecture · not an AWS deployment
WHO IT’S FOR
Build a provider-aware request path while keeping early usage visible.
Manage multiple model integrations through one application-facing API.
Inspect request health and configure practical controls per project.
Estimate usage and define a budget limit before spend surprises you.
EARLY ACCESS
No billing integration is implemented. Local demo usage is simulated; connecting an external provider uses your own account and may incur provider charges.
See availability and limitationsFAQ
Routiqel is an early-stage project. Product behavior and availability are described as they exist in this local MVP.
Browse documentationNo. Routiqel is an infrastructure layer that routes requests to a deterministic local adapter or an AI provider account you configure.
No. The local demo adapter works without provider credentials. It returns deterministic simulated responses.
No. The dashboard records estimates from available token counts and configured example rates. They are not invoices or guaranteed provider charges.
Streaming is not supported in this MVP. The API accepts non-streaming chat completion requests.
No. The local MVP runs on your machine with SQLite. AWS services are documented as a proposed production architecture only.
Configured credentials are encrypted by the local API before storage and are never included in browser responses. Production key management still needs a deployment review.
TAKE CONTROL OF YOUR AI INFRASTRUCTURE
Explore the simulated demo or contact us about the project.