🤖Chatbot Technical FAQ

General Questions

What AI models does Wonderchat support?

Wonderchat supports all major AI models, including the latest OpenAI (GPT) models, Anthropic (Claude), Grok, Gemini, Deepseek models and many more. You can select the model that best fits your needs based on accuracy, speed, and cost considerations.

Training & Data

What types of data can I use to train my chatbot?

You can train your chatbot using:

• Website pages (via URL crawling)

• PDF documents

• Word documents

• Video and audio files

• Manual text input

• Google Drive files

• Microsoft SharePoint documents

How often does the chatbot update its knowledge?

• Manual updates: You can trigger updates anytime through the dashboard

• Automatic crawling: Set up scheduled crawling for websites to keep content fresh (monthly, weekly or daily, depending on your subscription plan)

What does changing the Chatbot Context Length do?

Changing the Chatbot Context Length allows your chatbot to search through more data when looking for an answer to your user's question. This is particularly helpful for large and complex knowledge bases where an answer may be found in multiple parts of your knowledge base.

Is there a limit to how much data I can add?

Data limits depend on your subscription plan. Enterprise plans offer significantly higher limits for pages, files, and total knowledge base size.

Can the chatbot learn from conversations?

Yes, through the corrections feature, you can review chat logs and correct responses. The chatbot will remember these corrections for future conversations.

Integration & Deployment

How do I add the chatbot to my website?

You can embed the chatbot using a simple script that you can copy and paste into your website. We support the following formats for the chatbot in your website:

• A chat widget (floating button)

• A full chat screen (embedded in a page)

• A search bar

Can I integrate the chatbot with other tools?

Yes, Wonderchat supports integrations with:

• CRM systems (HubSpot, Salesforce via Zapier)

• Communication channels (Slack, WhatsApp, Facebook Messenger, Instagram, Discord)

• Support systems (Zendesk, Freshdesk)

• Automation tools (Zapier, Make)

Is there an API available?

Yes, Wonderchat provides a comprehensive API for:

• Sending chat messages

• Managing pages and content

• Exporting chat logs

• Managing tags and feedback

• Custom integrations

For more information, you can view our API documentation over at https://docs.wonderchat.io/api-reference/api-reference.

Performance & Accuracy

How accurate is the chatbot?

Our chatbots use a sophisticated retrieval and answer pipeline to maximize accuracy, with many customers reporting an over 70% resolution rate. Accuracy depends on:

• Quality and completeness of training data

• Selected AI model

• Proper configuration and corrections

• Regular updates to keep information current

What languages does the chatbot support?

Wonderchat supports 100+ languages. You can set up multilingual chatbots that automatically detect and respond in the user's language.

How fast does the chatbot respond?

Response times typically range from 1-5 seconds, depending on:

• Selected AI model

• Complexity of the query

• Amount of data to search through

Can the chatbot handle complex questions?

Yes, the chatbot can:

• Understand context and follow-up questions

• Handle multi-part queries

• Provide detailed explanations

• Reference specific sources

You can also set up detailed workflows to handle complex customer conversation flows.

Security & Privacy

Is my data secure?

Your data security and privacy is our top priority at Wonderchat.

• We are GDPR and SOC2 Type II compliant

• We have opted out of all data retention from our model providers, meaning that your chats are not used to train any underlying AI models from OpenAI or other providers.

• All data is encrypted in transit and at rest

• Servers are hosted in secure data centers

• Regular security audits and updates

Can I control what information the chatbot shares?

Yes, you can:

• Set specific response guidelines

• Exclude certain URLs or content

• Configure what information to collect from users

• Set up approval workflows for sensitive topics

Does the chatbot store user conversations?

Yes, Wonderchat stores your user conversations on our own secure database. Chat logs are stored for:

• Quality improvement and corrections

• Analytics and insights

• Customer support purposes

You can export/delete data your chatlog data as needed. We also support configuration of your own data retention policies on our Enterprise plan

Troubleshooting

Why isn't my chatbot answering correctly?

Common issues and solutions:

• Insufficient data: Add more relevant content

• Outdated information: Update your knowledge base

• Update your prompt: Update your chatbot role prompt to follow specific rules to answer questions more correctly.

• Wrong AI model: Try a more advanced model to ensure that the chatbot can answer complex questions and follow your instructions more closely.

• Need corrections: Review and correct responses

If your chatbot continues to answer questions incorrectly, please email us at [email protected]

The chatbot says it doesn't know something that's in my data

This could be due to:

1. Recent changes to your website may not have been indexed yet. Please trigger a manual recrawl of your site if recent information has updated

2. Content on the website or documents may not have been properly parsed. Please ensure that your documents are machine readable, or contact us at [email protected].

3. Conflicting information in different sources. Please ensure that multiple corrections have not been made for the same question with conflicting answers.

4. Conflicting prompt. Please ensure that your underlying chatbot role prompt does not have conflicting instructions that prevent it from answering

How do I improve chatbot performance?

1. Regularly update your documentation to ensure that the chatbot is not giving outdated information

2. Use a more advanced AI model to ensure that the chatbot follows instructions more closely

3. Use the corrections feature to fix mistakes

4. Remove unnecessary or redundant knowledge from the chatbot. This can sometimes confuse the AI model

5. Monitor analytics to identify knowledge gaps

6. Test with real user scenarios

7. Configure appropriate fallback responses

Customization

Can I customize the chatbot's personality?

Yes, you can:

• Set custom greeting messages

• Define response tone and style

• Add personality to responses

• Configure language and formality level

Can I add custom functionality?

Through chatbot tools and workflows, you can:

• Connect to external APIs

• Trigger custom actions

• Collect specific information

• Route to human agents based on conditions

How do I style the chatbot to match my brand?

We support the following customizations for your chatbot

1. Custom colors and themes

2. Logo and avatar customization

3. Name and tagline customization

4. CSS customization for advanced styling

5. Widget size and position

Please visit our documentation for more details

Can I remove the "Powered by Wonderchat" from my chatbot?

We do not currently offer this as an option for our retail plans. Please contact us at [email protected] if this is necessary.

---

Have more questions? Contact our support team or check our detailed setup guides for step-by-step instructions.