An old-style backend is where people find the screen, press the buttons and type every row themselves. In Vybev, that work moves to an AI agent. Just say it or type it: from revenue reports to customer care to data entry, the agent does it for you.
The security layer stays exactly where it was. The AI never holds the database keys; it works only through the actions the system already allows.
Agents per module, bound by ACL
Type “revenue by branch last month” or “10% off every drink until Sunday” and it is done — on your real numbers, and only within what that person is allowed to do.
A chatbot that guesses, with no idea who is allowed to see what.
A cashier asking for payroll gets nothing: the same rules as the screens.
Open the orders page → filter by month → sort by branch → export CSV → open a spreadsheet.
Want a discount applied to 40 products? Forty clicks, or a developer.
Every new question means a new report someone has to build.
200 stock lines from a delivery note? One cell at a time.
“Revenue by branch last month, worst three first.”
“10% off every drink until Sunday.”
“Reply to today's customers asking where their order is.”
“Enter this delivery note into stock.”
The agent never reads the database directly. It calls the system's own actions, carrying the role of whoever is asking, so it answers with real data and can only do what that person is allowed to do.
Reports, customer care, data entry, repricing a whole range: work that used to take a screen and dozens of clicks now takes one sentence. Your customers can ask too, on web and app alike.
The agent has no direct access to the database. Every read and write goes through the same API and permission layer the UI uses.
A cashier asking for payroll gets nothing. Permissions apply to the AI exactly as they apply to the UI.
If the AI enters something wrong, the FIELD rules on the server still stop it before bad data is saved.
From $59, lifetime. Free updates. Unlimited projects on your own server.