[{"data":1,"prerenderedAt":9},["ShallowReactive",2],{"docs-ai-en":3},{"title":4,"excerpt":5,"body":6,"seo_title":7,"seo_description":8},"AI agents","How the bot's AI personas are created: a name, a plain-language role prompt, and advanced model settings.","\u003Cp>The section where you create the bot's AI personas. An agent can be dropped into a scenario message so the subscriber can chat with it freely — until the completion condition set in the message's settings is met (the platform's own example: \"the agent will end the interaction after answering 5 of the subscriber's questions\").\u003C\u002Fp>\u003Ch2>Agent form\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cstrong>Agent name\u003C\u002Fstrong> — e.g. \"Company consultant\" (for your own navigation)\u003C\u002Fli>\u003Cli>\u003Cstrong>Agent role\u003C\u002Fstrong> — a system prompt in plain language: who the agent is, how it answers, what it recommends (the platform's example: \"You're a polite company assistant. Answer customer questions about our services. If a customer wants to book, recommend they tap the booking button.\")\u003C\u002Fli>\u003Cli>\u003Cstrong>Additional settings\u003C\u002Fstrong> — a collapsible block with finer model configuration (used less often, for non-standard scenarios)\u003C\u002Fli>\u003C\u002Ful>\u003Cp>One agent can be reused across several messages\u002Fscenarios — there's no need to create a new agent per scenario if the role stays the same. And how its conversation ends is configured not in the agent itself, but in the message that invokes it (see \"Completion conditions\" in the \"Scenarios\" topic).\u003C\u002Fp>","AI agents in TvoyBot: how to create and configure","How to create an AI agent in TvoyBot: naming it, writing its role prompt, and hooking it up to scenario messages.",1783936925304]