Our Own AI Gateway

Every AI call in Fedna AI — voice agents, text agents, qualification, summaries, Co-Pilot — now goes through our own AI gateway. This is infrastructure work you mostly won't see, but it makes the whole platform faster, more reliable, and more transparent.

Automatic failover

When an AI provider has a bad minute — rate limits, timeouts, 500s — calls automatically failover to a backup route. Your conversations don't break because one provider is having a rough day.

Per-organisation isolation

Every organisation has its own isolated gateway identity. Your usage, your limits, your metadata — completely separate from every other customer on the platform. This is the foundation for enterprise-grade controls coming soon (bring-your-own-keys, spend caps, model governance).

Per-event cost attribution

Every AI call is metered with full attribution — which agent, which customer, which channel, which conversation, which call. This powers the per-customer cost analytics you'll see surface in the Usage dashboard.

One place to manage models

Model choice and availability are controlled centrally. When we add a new model — or retire an old one — it propagates everywhere without you touching anything.

What you get out of it

• Fewer mysterious AI hiccups
• Faster recovery from provider issues
• Usage metering you can actually trust
• A platform that keeps pace as new models ship

The gateway is live for everyone today.