How it works
Your application submits a task with a deadline. Duale AI selects a model from the pool your team approved, runs the work, and returns a typed result.You submit the work
A task, the shape of the result you expect, a deadline, and the routing preferences for that piece of work. You name no model.
Duale AI runs it
It selects a model from the pool your team approved, calls the provider you contracted with, and can try another approved route before it returns. The deadline can be hours ahead, and the agent can read earlier content in its original form.
You get a typed result
Your application validates it before the business process uses it. Every tool you write runs in your own process, not on the platform.
What these pages cover
Security and hosting
Review security controls, managed-data hosting, subprocessors, certification status, and security contacts.
Pricing and access
How Duale AI charges for runtime work, what stays on the invoice from the model providers you contract with, and how to request access.