Your AI Agent

AI Providers & Models

Choose which AI model powers your agent and optionally use your own API account.

Overview

Your AI agent is powered by a large language model (LLM) — the brain that understands messages, decides which tools to use, and generates responses. The model you choose affects everything: how well the agent understands complex requests, how accurately it follows your instructions, how it reasons through multi-step tasks, and how it handles nuanced conversations.

By default, your agent uses the Platform Default provider, which requires no setup. If you have your own subscription to an AI provider, you can connect it for more control over model selection and access to the latest models.

AI provider options

Go to your instance settings and scroll to the Advanced section to see the provider options.

Instance Settings → Advanced

Platform Default

Shared platform key. No setup required.

Shared

Claude (Anthropic)

Uses your Claude subscription

Connected

OpenAI (GPT)

Uses your OpenAI subscription

Not Connected

Google Gemini

Uses your Google Gemini account

Not Connected
AI provider selection with model overrides

Platform Default

Uses the platform's shared AI key. No setup required — this is what you get out of the box.

This is the simplest option and works well for most use cases. The platform selects an appropriate model for your conversations.

OpenAI (GPT)

Use your own OpenAI subscription. This gives you access to models like GPT-5.3 Codex and GPT-5.2.

Claude (Anthropic)

Use your own Anthropic subscription. This gives you access to models like Claude Opus 4.6, Claude Sonnet 4.6, and Claude Haiku 4.5.

Google Gemini

Use your own Google AI subscription. This gives you access to models like Gemini 2.5 Pro, Gemini 2.5 Flash, and Gemini 3 Pro.

Connecting your own AI account

To use your own provider, you first need to connect it at the organization level:

  1. Go to the AI Providers page from the main dashboard sidebar
  2. Find the provider you want to connect (OpenAI, Claude, or Google Gemini)
  3. Click Connect
  4. Follow the authentication flow to link your account
  5. Once connected, the provider card shows a green Connected status

After connecting at the organization level, you can select the provider on any of your instances.

Selecting a provider for your instance

  1. Go to your instance settings
  2. Under Advanced, choose the provider you want to use
  3. If the provider isn't connected yet, you'll see "Not Connected" — click the link to go to the AI Providers page first
  4. Click Save

Model overrides

When using your own provider (not Platform Default), you can choose specific models for different tasks.

Available model selections

SettingWhat it controls
Main Chat AgentThe model that handles incoming messages and conversations
Sub-Agents (Delegate)The model used when your agent delegates tasks to specialized sub-agents
Scheduled TasksThe model used for scheduled messages and automated follow-ups

For each setting, you can either inherit the model from your organization's default or pick a specific model.

Reasoning effort

Some models support adjustable reasoning effort, which controls how much "thinking" the model does before responding:

  • Low — Faster responses, less detailed reasoning
  • Medium — Balanced speed and quality
  • High — More thorough reasoning, may take slightly longer

This setting only appears when your selected model supports it.

Which provider should I choose?

If you're just getting started, the Platform Default is the easiest option. You don't need to configure anything.

If you want more control, here's a general guide:

  • Claude (Anthropic) — Great for nuanced conversations, following complex instructions, and maintaining a natural tone. Good at being cautious and not making things up.
  • OpenAI (GPT) — Versatile and widely used. Strong for both conversational and analytical tasks.
  • Google Gemini — Strong multilingual support. Good if your customers communicate in multiple languages.

The best way to choose is to test — connect your preferred provider, send some test messages, and see which model gives the best responses for your specific use case.

Switching providers

You can switch providers at any time. Just go to your instance settings, select a different provider, and save. The change takes effect on the next incoming message.

Switching providers doesn't affect your conversation history or Knowledge Base — those stay the same regardless of which model is active.

Next steps

Now that your agent is configured, learn how to manage conversations:

  • Chat — View and manage your agent's conversations
  • Escalations — Handle conversations that need human attention