AI Providers: Connect LLMs to Power Your AI Agents
Connect OpenAI, Gemini, Claude, DeepSeek, OpenRouter, or any OpenAI-compatible LLM to your ChatbotX workspace to power your AI Agents.
Before an AI Agent can think and respond, you need to link your ChatbotX workspace to at least one AI Provider — the large language model (LLM) platform that supplies the intelligence. ChatbotX supports direct API integrations with all major LLM providers so you can choose the model that best fits your budget, accuracy requirements, and use case. You pay the provider directly based on your actual token consumption, with no markup from ChatbotX.Connecting an AI Provider gives you three key benefits:
Flexibility — pick any LLM that matches your conversational needs, from cost-efficient models to the most capable ones.
Automation — enable your Agent to study your Knowledge base and reply to customers automatically, 24/7.
Cost control — pay only for what you use, billed directly by the provider.
Navigate to Settings in the left navigation menu, then click the Integrations tab on the top toolbar. This page is where you add, manage, and remove all AI Provider connections for your workspace.
The most popular AI platform, providing highly intelligent and reliable responses across a wide range of tasks.API Key URL:https://platform.openai.com/api-keys
In ChatbotX, expand the OpenAI section under Settings > Integrations.
3
Paste your API Key and connect
Paste your copied API Key into the field and click Connect.
4
Configure Automated Responses
Set the Automated Responses option to match your preferred behavior:
Setting
Behavior
Enabled
Every inbound message triggers the AI model to analyze your knowledge base and reply automatically (Global Auto Reply).
Disabled
Global auto-reply is off. You can still use OpenAI by adding OpenAI actions manually inside individual Flow Builder flows.
2. Google Gemini
Features exceptionally fast processing speeds and a massive context window, making it well suited for large documents and multi-turn conversations.API Key URL:https://aistudio.google.com/app/apikey
Click Get API key and generate a new key. Copy it.
3
Add the key to ChatbotX
In ChatbotX, expand the Google Gemini section under Settings > Integrations, paste your key, and click Connect.
3. Claude AI
Anthropic’s Claude models are renowned for superior logical reasoning and professional, natural conversation quality.API Key URL:https://console.anthropic.com/settings/keys
Navigate to the API Keys section and generate a new key. Copy it.
3
Add the key to ChatbotX
In ChatbotX, expand the Claude AI section under Settings > Integrations, paste your key, and click Connect.
4. DeepSeek
Next-generation models that deliver high performance and strong language understanding at highly optimized API costs.API Key URL:https://platform.deepseek.com/api_keys
Select API Keys from the left menu and generate a new key. Copy it.
3
Add the key to ChatbotX
In ChatbotX, expand the DeepSeek section under Settings > Integrations, paste your key, and click Connect.
5. OpenRouter
A unified gateway that gives you access to hundreds of open-source and commercial AI models through a single API Key.API Key URL:https://openrouter.ai/keys
Create a new API Key in your key management settings. Copy it.
3
Add the key to ChatbotX
In ChatbotX, expand the OpenRouter section under Settings > Integrations, paste your key, and click Connect. Then select your desired LLM from the connected dropdown list.
6. OpenAI Compatible (Custom Providers)
OpenAI Compatible lets you connect any AI provider — including self-hosted models and third-party platforms — that exposes an OpenAI Chat Completions-compatible endpoint. This means you are not limited to the built-in providers; you can configure as many custom connections as you need.
1
Open the OpenAI Compatible section
Go to Settings > Integrations, expand the OpenAI Compatible section, and click + Add AI Provider.
2
Select a preset or choose Custom
In the Add AI Provider modal, choose a provider from the preset list (NVIDIA NIM, LM Studio, Heroku, Clarifai, NEAR AI Cloud, and others) or select Custom to configure your own endpoint.
3
Fill in the connection details
Configure the following fields:
Field
Description
Name
A display label to distinguish this provider (e.g., “LM Studio Local”, “Ollama Server”).
Base URL
The API endpoint address for the provider. See the reference table below.
Default model
The default model name to use (e.g., gpt-4o-mini, llama-3.1-8b).
API Key
Authentication key from the provider. Some local services like LM Studio or Ollama may not require one.
AI Auto Reply
Enable to use this provider for the global auto-reply feature.
Enabled
Toggle on to activate this provider connection.
4
Save the connection
Click Confirm to save and activate the provider. You can add multiple OpenAI Compatible providers simultaneously — each appears as a separate option when configuring AI Agents.
Once you have linked your API Keys in Settings, configure the specific model and parameters for each Agent individually.
1
Open AI Agents
Go to the AI Agents menu in the left sidebar.
2
Select or create an Agent
Click an existing Agent to edit it, or click + Create AI Agent to build a new one.
3
Configure the model parameters
In the Agent configuration form, set the following parameters in the right panel:
Parameter
Description
Model selection
Choose the specific LLM version for your connected provider (e.g., Gemini Model, OpenAI Model, Claude Model, DeepSeek Model, OpenRouter Model, or any OpenAI Compatible provider you added).
Temperature
Controls the creativity and randomness of responses, on a scale from 0 to 1. Lower values (0.2–0.4) produce factual, precise replies ideal for customer support. Higher values (0.7–0.9) produce more expressive, creative language ideal for content generation.
Max Tokens
Limits the maximum length of each response (e.g., 2048). Helps prevent overly verbose outputs and keeps your API costs predictable.
4
Save the configuration
Click Confirm (or Save) to apply the model configuration to the Agent.
Each AI Agent can be assigned its own model and parameters independently. This lets you run a cost-efficient model for simple FAQs and a more capable model for complex sales conversations at the same time.
⌘I
Assistant
Responses are generated using AI and may contain mistakes.