ResonAi Gateway

01 How it works

Apps ask for a tier. The gateway picks the model.

Nothing in your apps names a vendor. That one rule is what lets you change models, cap spend and prove where data went, without rewriting anything.

Proposal writer tier: deep

Support chatbot tier: fast

PMO status agent tier: standard

HR policy assistant tier: ca-standard

RESONAI GATEWAY

A key per team or app, revocable on its own
A model allowlist per key, including a Canada-only lock
A hard budget per key: at the limit, calls stop
Fallback to a second vendor if the first is down
Every call logged: who, which model, tokens, cost
Prompts and replies are not stored

Claude Anthropic

GPT OpenAI

GPT in Canada East Azure

Others as needed open-weight

02 Try it

The gateway console.

Pick who is calling and which tier they want. Then try a vendor outage. This simulation follows the behaviour we measured on the real gateway.

gateway console · simulation

1 · Calling as

2 · Asking for tier

Simulate an outage at the primary vendor

Pick a key and a tier, then send.

Audit log

Time	Key	Tier	Answered by	Result	Tokens	Cost	Content

The keys in this demo

ops-teamall tiers · CA$100/moDay-to-day automations. May use any tier except Canada-only.

hr-canadaca-standard onlyHandles personal information. Locked to a model that processes in Canada. Anything else is refused.

support-botfast only · CA$40/moThe website chatbot. Cheap, quick model only, so a busy week can't run up the bill.

contractor-07budget reachedA contractor's key that hit its limit. Calls stop until an owner raises it.

03 Change models without changing apps

Swap the model behind a tier.

Pick the model that sits behind "standard". Watch the app code on the left. It does not change.

// inside your PMO status agent
const reply = await gateway.chat({
  model: "resonai-standard",
  key: "ops-team",
  tags: ["workflow:pmo-status"],
  messages
});

Behind "standard" today

Claude Sonnet

0 lines of app code changed

We only move a task to a cheaper model after it passes a quality test on that task. In our own test, the fast tier was four times cheaper but mislabelled the job type on 2 of 3 sample estimates, so that agent stayed on standard.

04 The bill, per workflow

See what each workflow actually costs.

Every call is tagged with its workflow and agent. This is one real run of our five-agent estimating workflow, measured through the gateway.

Usage report · workflow: estimator

key: demo-client · tier: standard · content not stored

5 agents · 31,817 tokens · one full estimateCA$0.21

Model cost only, measured 2026-09-30 and converted from US dollars at the Bank of Canada rate for 2026-09-29 (1.4188). It does not include ResonAi's fees.

The plain truth on data residency. Storing your data in Canada is easy. Running the model inside Canada limits your choice of model today: Canada-only processing currently means specific OpenAI models in Azure's Canada East region. Claude can keep your data stored in Canada, but its processing runs outside Canada. The gateway lets you lock only the workflows that need it.

Want a gateway in front of your AI?

We set it up in your cloud or ours, move your existing AI tools behind it, and hand you the keys and the reports.

Book a 30-minute call See it in the enterprise stack Back to the Demo Lab