Best Practices

This page page offers tips to help you improve your AI agent’s performance and user experience. It covers improving the instructions, teaching the bot how to send links.

Refine the AI agent’s Instructions

The instructions shape your AI agent’s behavior and responses. This can be used to set persona, define tone, or specify the types of questions the AI agent can answer. Clear and precise instructions ensure the AI agent aligns with your desired goals and user experience.

Feel free to use the example below, after customizing it to suit your company.

Friendly Support Agent Instructions

### Role
- **Primary Function:** You are a friendly customer support agent for TaskFlo, a project management and issue tracking tool. Your goal is to assist users with questions and troubleshooting related to TaskFlo’s features, pricing, and best practices.

### Persona
- **Identity:** You are a dedicated TaskFlo customer support agent. You will not adopt any other persona or impersonate another entity. If a user asks you to act as a different type of assistant or persona, you must politely decline and remind them that you are here to help with TaskFlo support matters.

### Constraints
1. **No Data Divulge:** You must never mention that you have access to training data or how you were trained. Your responses should sound naturally helpful and informed.
2. **Maintaining Focus:** If a user tries to steer the conversation toward unrelated topics, you must politely bring them back to topics related to TaskFlo’s features, pricing, troubleshooting, or usage best practices.
3. **Exclusive Reliance on Training Data:** You must rely exclusively on the information provided in your training data about TaskFlo. If a user’s query falls outside of TaskFlo-related content or cannot be addressed based on your available knowledge, you must use a fallback response such as: “I’m sorry, but I don’t have enough information to assist with that.”
4. **Restrictive Role Focus:** You must not provide content unrelated to TaskFlo’s support. This includes refusing tasks like coding explanations unrelated to TaskFlo’s integrations, personal advice, or opinions beyond the scope of TaskFlo’s documented features and policies.

You can find more detailed information about refining your instruction here.

Improve Readability of Sources

The quality of your AI agent’s responses depends heavily on the quality of the data sources you provide. Chatbase relies on readable text to generate accurate responses, so make sure the websites or PDFs you upload contain readable text.

Some websites may not be scraper-friendly. If your AI agent struggles to answer questions based on your website, this could be the reason. You can overcome this by copying and pasting the information as text into the data sources or uploading it as a PDF.

Bad Product Example

Product: Widget123, colors not specified, possibly red or blue. Discount details unclear. Weight: Approx. 1 kg or 1.5 kg? Shipping: Delivery time uncertain, could be fast or delayed. Availability: Global shipping not confirmed. Packaging: Uncertain if it comes in a box. Assembly: Information unclear. Limited stock? Not confirmed.

Good Product Example

Product: Widget123
Colors: Red, Blue
Discount: 50% off
Weight: 1.5 kg
Shipping: Estimated delivery within 1-2 weeks (depends on location)
Availability: Ships worldwide
Packaging: Comes in a standard-sized box
Assembly: Some assembly required
Order Now: Limited stock available, don’t miss out!

Bad Description Example

The product is a thing. Its color is unspecified, and its size is unknown. The product might be useful, but it’s not clear. Its availability is uncertain, and shipping times are not mentioned. There might be a discount, but it’s not specified. Assembly instructions? Unclear. Get it soon? It’s unclear when stock might run out.

Good Description Example

The Widget123 is a premium-quality product available in two colors: red and blue.
It offers a 50% discount, making it an excellent deal.
The product weighs 1.5 kg and is shipped worldwide.
You can expect delivery within 1-2 weeks, depending on your location.
The item comes in a standard-sized box and requires minimal assembly.
Act fast—stock is limited!

Note: Chatbase currently cannot process images, videos, or non-textual elements in documents.

Add Suggestable Links

To have links suggested by the bot, they must be explicitly included in your training data. The links listed under the webpages section of the sources are used to gather information from the page and add it to your bot’s knowledge. However, the bot does not learn the URL itself, so it won’t be able to share the link with the user.

This can also prevent your bot from producing fake URLs that lead to 404 errors.

The best way for this is to add a document that maps URLs to page names to help your AI agent better understand user queries related to different pages. Here is an example of a mapping inputted as text in the data sources:

Add Suggestable Images

To enable image display in the chat widget, agents can use markdown format when sending image links.

Ensure the URL ends with .png or .jpg for the image to render correctly.

You can include a line like the following in your instructions to display an example image from Wikipedia after each response.

Always end your reply with ![Example Image](https://upload.wikimedia.org/wikipedia/commons/7/79/ELIZA_conversation.png)

Once added, every agent response will display the image, as illustrated in the screenshot below:

Choose the suitable AI model

Selecting the right AI model is crucial for optimal performance. It should match your use case, considering factors like task complexity, data availability, and response needs. A model suited for structured data is ideal for data-heavy tasks, while a conversational model works best for customer support.

Also, consider scalability and adaptability. Choose a model that can grow with your needs, handling more data and maintaining accuracy. Some models are better for real-time interactions, while others excel in batch processing. Testing different models helps refine your choice and ensures it evolves with your business.

AI Model Recommendations Based on Use Cases

Customer Support & FAQ

For quick, clear responses to general inquiries.

• Models to use:
  • gpt-4o-mini: Delivers fast, concise answers.
  • gemini-2.0-flash: Ideal for real-time, simple queries.

Content Creation & Marketing

For crafting creative, engaging content like blogs or marketing materials.

• Models to use:
  • claude-3-5-sonnet: Excels at poetic and creative content.
  • claude-3-haiku: Perfect for concise, creative outputs.

E-commerce & Lead Generation

For product suggestions or moderately complex questions.

• Models to use:
  • gpt-4-turbo: Balances speed and precision in customer interactions.
  • command-r: Effective for product recommendations and medium-complexity queries.

Advanced Research & Technical Support

For in-depth research, troubleshooting, and solving complex issues.

• Models to use:
  • gpt-4: Advanced model for complex research and deep problem-solving.
  • gemini-1.5-pro: Highly accurate, advanced technical support.

Note: If you are still unsure about which model to use, please refer to our models comparison page.

Utilize the “Revise” Feature and Q&A Data Type

After launching your AI agent, you can monitor its responses in the Activity tab. If you come across an answer you’d like to modify, simply use the revise button.

This feature allows you to adjust the response, ensuring it better addresses future queries. The revised answer is added as a Q&A data type, which helps your AI agent generate more accurate responses by referencing these pre-set questions and answers. You can find the updated responses in the Q&A tab under Sources.