How to build an AI agent that can act in your app
In DYPAI, you can create an AI agent inside your app and give it specific backend endpoints as tools. A user can then ask the agent a question or request an action in plain language. The agent chooses among the tools you attached, calls an endpoint in the same project and uses the result to answer. You decide which operations it can reach and how those endpoints protect data.
This is different from asking an AI to build an app. Once your app is running, its own agent can help people use it: find a booking, check stock, update a record or start a defined workflow. The agent documentation covers the underlying feature; this article shows the product decision and a practical path to building one.
An example: a booking assistant with real tools
Imagine a booking app with two existing endpoints:
list_available_slotsaccepts a date and returns times the signed-in user may book.create_bookingaccepts a slot and creates a reservation after the app has verified the user's approval.
You mark both endpoints as tools and attach them to a booking_assistant agent. A customer asks, “What is available on Friday afternoon?” The agent calls list_available_slots and returns actual times from the app. After the customer chooses a time and explicitly approves the booking, the agent calls and reports the result returned by that endpoint.


