🧠 Pricelist AI
Pricelist AI is the artificial intelligence system integrated with Pricelist Server that allows managing, testing, and deploying customized intelligent assistants. The module’s goal is to provide a comprehensive interface to configure AI behavior, monitor its responses, and connect it to WordPress through a dedicated plugin, offering end users an interactive chatbot directly on the client’s website.
ℹ️ Info
This section shows the main license data and the active status of Pricelist AI. It also contains the link to download the WordPress plugin that allows integrating the chatbot into the client’s site.
📋 Main Fields
• License ID → Unique license identifier.
• Valid Until → License expiration date.
• Status → Current status (Active / Inactive).
• Download Plugin → Link to download the WordPress integration plugin.
🔧 Using the WordPress Plugin
1. Access the WordPress admin panel.
2. Open Plugins → Add New.
3. Select Upload Plugin and import the downloaded file.
4. Click Install Now and then Activate Plugin.
5. After activation, configure the connection parameters to Pricelist Server to make the chatbot available on the website.
The chatbot will then be able to respond to questions about the price list and products published online, using information synchronized with Pricelist Server.
⚙️ AI Setup
In this section, the operational parameters of the artificial intelligence are defined. You can adjust the “temperature” and choose the language model to use for responses.
📋 Available Parameters
• Temperature → Numeric value that controls the variability of responses.
• Low values (0.1–0.4) produce more precise and consistent responses.
• Medium values (0.5–0.7) balance accuracy and creativity.
• High values (>0.8) make responses more variable and generative.
• Model Selection → Allows you to choose the active AI model.
• gpt-3.5-turbo → fast model, recommended for testing and prototyping.
• gpt-4 → stable and precise model, ideal for structured responses.
• gpt-4o → model optimized for performance and advanced integrations.
💾 Operational advice: for the production environment, it is recommended to maintain a Temperature of 0.7 and use the gpt-4o model.
🧾 Prompt Viewer
View the list of prompts available in the system, divided by environment (DEV and PROD).
Each row shows the total number of prompts, distinguishing between reserved and non-reserved ones.
• Reserved Prompts → prompts with limited access, such as sensitive data like prices or commercial terms, visible only to registered users.
• Non Reserved Prompts → public prompts usable by any user.
This section allows you to verify that the active prompts are consistent with the AI environment configuration.
🧪 Test Unit
Allows you to perform functionality tests on artificial intelligence and verify the quality of responses.
You can create new tests, execute them, and manually evaluate the results.
📋 Main Features
• AI Quality Score → overall quality score of responses.
• New Test → opens the Add Test Case window to create a new custom test.
• Run Test → executes the test and generates the AI response.
• Evaluation → manual selection of judgment (Good / Bad).
• Expected / AI Reply → comparison between expected response and generated response.
🧩 Add New Test
Allows you to manually add a new test by filling in the following fields:
• Topic → selection of the test topic or category.
• User Input → test message or question to send to the AI.
• Expected Output (optional) → expected response, useful for comparing the test outcome.
Once the fields are completed, press Save to add the test or Cancel to cancel the operation.
💡 Tip: use this section to test specific prompts before making them public in a production environment.
📊 RAG Status
Manages the loading and verification environments for documents in the RAG (Retrieval-Augmented Generation) system. The environments are divided into:
• DEV Environment → development environment for testing and updates.
• PROD Environment → production environment, used by the active chatbot.
🔁 Button “Put DEV in Production”
Promotes the content from the development environment to the production environment, making new configurations or updated documents operational.
🪪 Identity
Contains the main prompt that defines the identity and behavior of the AI assistant. Here, tone, language, response rules, and interaction limits are specified.
📋 Main Fields
• Assistant Identity → description of the assistant’s role and function.
• Language Behavior → language management rules based on user input.
• Behavioral Instructions → behavioral instructions on tone, respect, and accuracy.
💾 Press Save to apply changes and update the model’s identity.
📜 AI Logs
Records all interactions and tests performed with Pricelist AI, showing:
• Session ID,
• User message,
• AI response,
• Topic of the prompt used,
• Date and time of execution.
Each log allows detailed analysis of the model’s behavior, including the prompts selected to generate the response, thus providing in-depth control over accuracy and consistency.
🧠 Operational Tips
• Always use the DEV environment for testing and promote to PROD only stable versions.
• Maintain a constant temperature to ensure uniformity in responses.
• Regularly monitor logs to identify any errors or inconsistencies in prompts.
• Periodically update the assistant’s identity to align with tone and company policies.
• Before publishing changes, always verify results through the Test Unit section.