Bot settings are essential configurations that must be properly set up before running the bot. Ensuring these settings are correctly configured is crucial for the bot’s optimal performance and functionality, as it directly impacts its efficiency and the accuracy of its tasks.

Simultaneous Bot Testing Feature in AptlyStar

In AptlyStar bot settings, you can directly configure the bot while simultaneously viewing the bot beside the settings options. This feature provides several key benefits:

  • Real-Time Testing: You can test the bot immediately after making changes, ensuring quick validation of updates without switching between screens.

  • Streamlined Workflow: It eliminates the need to navigate back and forth, saving time and improving efficiency during bot customization.

  • Improved Accuracy: With instant feedback, you can fine-tune responses or behaviors on the spot, reducing errors and enhancing the user experience.

  • Enhanced Productivity: By combining settings and testing in a single interface, this feature enables faster iteration and deployment of a well-optimized bot.

Please refer the image below:

_____________________________________________________________________________________________

Configuring Bot Settings Part-1:

Please refer the below image.

  • Go to the “Settings” section inside your bot (1). You can change the project (bot name) (2).

  • Ensure that the correct chat model (3) & embedding model are selected (4).

Review all settings to confirm they align with your requirements and finally save the settings.

Configuring Bot Settings Part-2:

Test the bot with Prompt with Context (1): The “Prompt with Context” ensures the AI answers questions using only the provided context. If unsure, it responds with “I don’t know” rather than guessing. For unrelated questions, it politely states it’s limited to context-specific answers.

Test the bot with Prompt without Context (2): Meaning: This prompt type is used to reformulate follow-up questions into standalone ones, making them clear without relying on previous conversations. It’s useful for ensuring questions are self-contained.

Test the prompt with context (1), test the prompt without context (2), insert the query (3), and view the responses displayed on the screen in the bot settings sidebar (4).

Prompt Library (3): The Prompts Library provides a curated list of pre-created prompts that you can select and customize as needed.

  • Click on Prompt Library (1).

  • Choose a prompt from the pre-created library list.

  • Once selected, you can either edit the prompt to customize it or use it as is.

Configuring Bot Settings Part-3:

  • In AI bot applications, temperature is a parameter that controls the randomness of the bot’s responses.

  • Higher temperature (e.g., 0.8-1.0) makes answers more diverse and creative but less predictable. Lower temperature (e.g., 0-0.3) makes responses more focused, deterministic, and repetitive.

  • This setting helps balance between creativity and accuracy based on your application’s needs.

  • Please refer below image. Drag the slider to adjust the temperature of the bot.

Configuring Bot Starter Questions:

The Starter Questions section allows you to define up to four user-focused questions that are most relevant and practical for your bot users. These questions are designed to help guide users and provide quick access to common topics or actions.

How to Configure Starter Questions: Please refer the image below:

Navigate to Bot Settings (1) and click on Starter Questions (2), then define up to four concise, user-focused queries, add them (3), and save (4).

The configured questions will appear in your bot’s Playground. Users can click directly on these questions to initiate interactions, ensuring a seamless and engaging experience.

Query Parameters:

Query parameters allow you to pass specific data to your bot to customize its behavior or responses based on user input or contextual information. This can be useful for tailoring interactions to meet specific user needs or scenarios.

How To Add Query Parameter:

Please refer the below image:

  • Navigate to (1) Bot Settings, click (2) Add Query Parameter, enter your desired parameter name (3), update the (4) prompt by inserting relevant parameters and save changes.

  • Navigate to the Playground, the query parameters will be appended above the questions area, reflecting any parameters that have been added. You can start asking queries without needing additional details.

The bot will process these parameters and adjust its responses accordingly. By adding query parameters, you enhance the bot’s flexibility and enable personalized, context-aware interactions for users.

Semantic Search Similarity Scoring in AptlyStar:

The scoring system evaluates the relevance of search results in relation to the user’s input, using advanced AI algorithms. Results are assigned similarity scores to indicate their relevance, where higher scores represent more accurate and precise matches. The system classifies matches into thresholds:

Considering All Documents: Option to analyze all the documents or knowledge base, ensuring no relevant information is missed.

  • ≥ 0.2: Broad contextual relevance, capturing less related matches.

  • ≥ 0.5: Moderate relevance, identifying contextually related results.

  • ≥ 0.7: High relevance, delivering highly precise and accurate matches.

This scoring mechanism ensures users can easily identify the most relevant results for their needs.

Steps To Test Accuracy With Semantic Search Similarity Feature:

To switch the semantic score, click on Bot Settings (1) and select your choice of semantic search similarity score from the dropdown (2).

In the image below, we have switched between scores 0.2 and 0.7. Notice the difference and observe the impact on accuracy.