Dragwyb Click to Chat’s AI Assistant uses a selected AI provider and model to generate visitor replies. You can configure provider credentials, the system prompt that guides the assistant, response parameters, and request limits from the WordPress admin. This overview links to the existing setup guides for each task and distinguishes the usage controls available in the plugin from charges billed by your AI provider.
Open the AI Assistant settings
In WordPress admin, open Click to Chat → AI Assistant. The AI Assistant screen includes the Chatbot Settings and AI Engine & Prompt sections, along with Knowledge Base, Chat Sessions (when chat saving is enabled), Chat Preview, and Error Logs. The settings and available sections can vary with configuration and installed version.
Connect an AI provider
In AI Engine & Prompt, use AI Providers & API Keys to configure a provider credential, select the active provider, and choose an available model. The inspected version’s provider cards mark OpenAI and Google Gemini as available without the Pro add-on. Anthropic Claude, OpenRouter, Groq, DeepSeek, and Local Ollama are displayed as Pro-only provider options in this interface. Provider and model availability may also depend on the API account and current provider API.
Keep API keys private and configure them only in the intended admin interface. A provider key being saved does not, by itself, prove that requests will succeed; send a controlled test message and check the plugin’s error logs and the provider’s account status if it fails.
Configure instructions and response tone
Use AI Engine & Prompt → System Persona & Instructions to edit the Custom ChatBot System Prompt. The prompt defines the assistant’s role, scope, response style, and boundaries. The inspected interface does not provide a separate dedicated tone selector in this section, so tone is primarily expressed through the instructions you write.
The same screen provides Chat Accuracy / Temperature, with a range of 0.0 to 2.0 and a default of 0.7, and Max Tokens (Response Length), whose interface accepts values from 100 to 8,192 and defaults to 500. Lower temperature values generally make responses more predictable; higher values permit more variation. Actual behavior also depends on the selected provider and model.
Control usage and request limits
Open Chatbot Settings → Advance → AI Usage & Cost Controls to enable or disable the usage controls and configure the visitor daily message limit and the site’s monthly request budget ceiling. In version 1.2.0, the defaults are 50 messages per visitor per day and 5,000 AI requests per site per month. A monthly ceiling of 0 means unlimited monthly requests. The separate Rate Limit (Requests per minute per visitor) field defaults to 20.
When budget email alerts are enabled, the plugin can send threshold notifications when the monthly request count reaches 80% and 100% of the configured ceiling. You can provide an alert email address or leave it blank to use the WordPress administrator email. The dashboard’s usage meter reports request counts and estimated tokens; it does not calculate the provider’s actual billed currency amount. Review your provider account for real charges.
Test changes before relying on them
- Confirm that the AI Assistant module is enabled and that at least one supported provider is configured.
- Send a short test message and confirm the expected provider/model is active.
- Test representative visitor questions against the system prompt, including a question the assistant should not answer from unsupported information.
- Review request limits and usage alerts using test traffic where possible. Remember that request-count ceilings are not dollar-denominated billing caps.
- If a response fails or behaves unexpectedly, review the configured model, provider status, prompt, knowledge-base retrieval if used, and the plugin’s Error Logs.
These are recommended checks, not a claim that live provider requests, email delivery, or billing behavior have been tested on your site.