General Questions

General FAQ

Find answers to the most frequently asked questions about Chatsby. If you do not find what you are looking for here, check the Billing FAQ and Security FAQ, or contact our support team at [email protected].

What is Chatsby?

Chatsby is a SaaS platform that enables businesses to build, train, and deploy AI-powered customer support agents. You provide your own data --- website content, documents, FAQs, product catalogs --- and Chatsby creates an intelligent agent that can answer questions, capture leads, assist with onboarding, and handle first-line support. These agents can be embedded on your website as a chat widget, deployed on WhatsApp, connected to Slack for team notifications, and integrated with tools like Zapier and Zendesk.

How is Chatsby different from ChatGPT or other general-purpose chatbots?

General-purpose chatbots like ChatGPT are trained on broad internet data and respond based on general knowledge. Chatsby agents are trained specifically on your data and respond only based on the information you provide. This means:

• Accurate, on-brand responses --- Your agent answers questions using your documentation, policies, and product information, not generic internet knowledge.
• No hallucination about your business --- The agent is constrained to your knowledge base and will not fabricate information about your products or services.
• Customizable behavior --- You control the tone, personality, and scope of responses through system prompts and settings.
• Built for deployment --- Chatsby includes embedding, integrations, analytics, lead capture, and escalation workflows out of the box. ChatGPT does not.

What AI models power Chatsby agents?

Chatsby leverages state-of-the-art large language models from leading AI providers, including OpenAI. Depending on your plan and preferences, you can choose between different models to balance response quality, speed, and cost. Model options are available in your agent's Settings tab. We continuously evaluate and add new models as the AI landscape evolves.

How accurate are the responses?

Accuracy depends on the quality and completeness of your training data. When your knowledge base covers a topic thoroughly, the agent will provide highly accurate responses. Best practices for maximizing accuracy include:

• Providing comprehensive, well-structured training data
• Covering common customer questions explicitly in your data sources
• Regularly reviewing conversation logs and updating your knowledge base to address gaps
• Using the system prompt to guide the agent's behavior and tone

The agent is designed to acknowledge when it does not have enough information to answer a question, rather than guessing or fabricating a response.

What types of data can I train my agent on?

Chatsby supports multiple data source types:

You can combine multiple data source types for a single agent. For example, upload your product catalog as a PDF, add your FAQ page URL, and manually enter your return policy text.

How long does training take?

Training time depends on the volume of data:

You will see a progress indicator in the dashboard while training is in progress. Your agent remains available during retraining --- it continues to use the previous knowledge base until the new training is complete.

Can the agent handle multiple languages?

Yes. Chatsby agents can understand and respond in multiple languages. The underlying AI models support over 90 languages. The agent will typically respond in the same language the visitor uses. For best results:

• Provide training data in the languages your customers use.
• If you serve a multilingual audience, include translations of key content in your data sources.
• You can also use the system prompt to instruct the agent to always respond in a specific language, regardless of the visitor's language.

Is there a limit on conversations?

Yes, each Chatsby plan includes a monthly message quota. A "message" is a single exchange --- one visitor message and one agent response. When you reach your plan's message limit, behavior depends on your plan and settings. See the Billing FAQ for details on quotas, overage handling, and plan upgrades.

Can I use Chatsby for internal (employee) use?

Absolutely. Many organizations use Chatsby to build internal knowledge base assistants. Common use cases include:

• Employee onboarding --- Train an agent on your company handbook, policies, and procedures.
• IT help desk --- Provide instant answers to common IT questions (password resets, VPN setup, software access).
• Sales enablement --- Help sales reps quickly find product specifications, competitive comparisons, and pricing guidelines.

For internal agents, set the agent to Private and use API authentication to restrict access to authorized users within your organization.

Does the widget work on mobile?

Yes. The Chatsby chat widget is fully responsive and works on all modern mobile browsers, including Safari on iOS and Chrome on Android. The widget automatically adapts its layout to smaller screens, and the chat interface is touch-friendly with appropriately sized input fields and buttons.

Can I have multiple agents?

Yes. You can create multiple agents within a single Chatsby account, each trained on different data sources and configured with different settings. This is useful for:

• Different products or brands within the same organization
• Separate agents for different departments (support, sales, HR)
• Testing new training data on a staging agent before updating your production agent

The number of agents available depends on your plan. See the Billing FAQ for plan details.

How do I improve response quality over time?

Chatsby provides several tools for continuous improvement:

1. Review conversations --- Use the Conversations tab in your dashboard to read real interactions and identify where the agent falls short.
2. Update training data --- Add new data sources or update existing ones to fill knowledge gaps.
3. Refine the system prompt --- Adjust your agent's system prompt to improve tone, scope, and response style.
4. Monitor analytics --- Use the Analytics tab to track satisfaction ratings, conversation volume, and common questions.
5. Use Actions --- Set up Actions to handle specific scenarios that require structured responses or data collection.

What happens when the agent does not know the answer?

When the agent does not have sufficient information in its knowledge base to answer a question, it is designed to respond honestly rather than guess. Typical behavior includes:

• Acknowledging that it does not have the information
• Suggesting the visitor rephrase the question or contact human support
• Triggering an escalation to a human agent (if configured via Slack or Zendesk integrations)
• Capturing the visitor's contact information so a team member can follow up

You can customize this fallback behavior through the system prompt and escalation settings.

Can I see what users are asking?

Yes. The Conversations tab in your agent dashboard provides a complete log of every conversation, including:

• Full message transcripts
• Visitor information (name, email, location --- if captured)
• Timestamps and duration
• Satisfaction ratings
• Page URL where the conversation started

You can also export conversation data for analysis and reporting.

Is there a free plan?

Yes. Chatsby offers a free tier that allows you to create an agent, add data sources, and test the chat experience. The free plan has limits on monthly messages and available features. See the Billing FAQ and our Pricing Page for complete details on all plans.

How do I get support?

Chatsby offers multiple support channels:

Enterprise plan customers receive priority support with dedicated account management and faster response times.