AI Assistants

LeadsWithDemos AI Assistant Overview

LeadsWithDemos AI Assistants leverage the power of OpenAI’s advanced AI models, integrating OpenAI’s LLM capabilities with custom knowledge and tailored instructions.

With the LeadsWithDemos AI Assistant, you can create unique AI assistants that seamlessly integrate with LeadsWithDemos popups and forms functionality to effectively capture leads. Furthermore, these AI assistants can also generate SEO-optimized content for your documentation, enhancing your overall content strategy.

LeadsWithDemos AI Assistant Types

Currently, in LeadsWithDemos, you can create two types of AI assistants:

SEO Assistants

These AI assistant types allow you to create content from an SEO perspective. They are ideal for generating documentation for your SaaS, and also allow you to translate the content in just a few steps.

To view the SEO Assistant details, navigate to AI Assistants > Select the Relevant SEO Assistant Type > Assistant Details.

Other Assistants

This category includes chat-based assistants that are perfect for lead capturing, website embedding, and knowledge sharing. They provide you with relevant and accurate information about your products or services.

To view the Other Assistant details, navigate to AI Assistants > Select the Relevant Other Assistant Type > Assistant Details.

Assistant Details Page

On the Assistant Details Page, you will find the following settings:

Name

The “Name” field on the Assistant Details Page allows you to assign or update a unique identifier for your assistant. This name serves as a reference, making it easier to identify and manage your assistant within the LeadsWithDemos platform. You should choose a descriptive name that reflects the assistant’s purpose or functionality; it has a maximum limit of 45 characters.

AI Model

The “AI Model” field on the Assistant Details Page allows you to select the specific AI model that your assistant will use from the dropdown list. This selection enables you to customize the assistant’s performance based on your needs, ensuring it delivers the desired outcomes. It is essential to choose an AI model that aligns with the intended functionality of your assistant for optimal results.

It is recommended to use the latest version of OpenAI’s models for the best performance.

Instructions

The “Instructions” text area on the Assistant Details Page allows you to provide specific guidance or directives for your assistant. These instructions help shape how the assistant interacts with users and performs tasks, ensuring that it aligns with your goals and expectations.

It is important to write clear and concise instructions that effectively communicate the desired behavior and outcomes for the assistant, enabling it to deliver optimal performance.

Tools

The “Tools” section on the Assistant Details Page includes options that enhance the capabilities of your assistant. This section features the Code Interpreter and File Search tools.

File Search

This option allows you to select specific vectors available from your organization settings, empowering your assistant to retrieve relevant and accurate data more effectively.

By utilizing the File Search tool, you can tailor your AI assistant with specific knowledge about your organization or unique custom knowledge.

To see a vector available in the dropdown list, you need to create it by navigating to AI Assistants > Settings > Vector.

Force File Search

The “Force File Search” option enables your assistant to prioritize the retrieval of the most relevant files, providing a more efficient and precise response to user queries. This feature is particularly useful in scenarios where quick access to specific information is crucial.

It is an ON/OFF toggle. When set to off, the assistant can consider retrieving answers from custom knowledge (vector). When set to ON, the system forces the assistant to retrieve information only from the custom knowledge (vector).

Temperature

The “Temperature” setting on the Assistant Details Page controls the randomness of the assistant’s responses. A lower temperature value (closer to 0) results in more deterministic and repetitive outputs, while a higher temperature value introduces greater variability and creativity in the generated responses. Adjusting the temperature allows you to tailor the assistant’s behavior to suit different contexts or preferences, enabling you to choose between precision and creativity in the assistant’s interactions.

We recommend leaving it at the default setting, which is a temperature of 1.

Top_p

The “Top_p” setting, also known as nucleus sampling, influences the diversity of the assistant’s responses by controlling the range of options considered during text generation.

Setting Top_p to a value of 0.5 means that the assistant will consider the top 50% of likelihood-weighted options when forming a response, allowing for a balance between consistency and variety. By adjusting the Top_p value, you can customize the assistant’s output to be more diverse or focused, depending on your needs.

We recommend leaving it at the default setting, which is a Top_p of 1.