Now, let’s shape how the bot behaves in the wild. AI training involves defining how it should answer questions, with a dash of personality and expertise.

There are two sections here. They are, ChatGPT Setting and Train AI.

ChatGPT Settings

Here’s where you make the AI smart enough to work for you:

1. ChatGPT Settings: Click Save Settings to lock in the changes when you’re done tweaking.

2. ChatGPT Training Type: You have options here:

• Prompt-Based: This is where you provide specific instructions for responses. It’s like giving the bot a script.
• Fine-Tune: More detailed customization, this trains the AI with your own data, making the responses feel more natural and tailored.

3. Select Open AI Model: Pick the OpenAI model that works best for you.

• GPT-3.5 for basic responses.
• GPT-4 for deeper, more complex interactions.
• GPT-4 Turbo for faster, more efficient responses.

4. Open AI API Key: To link the plugin with OpenAI, enter your API key here. Obtain it by signing up with OpenAI and generating the key in your account.

5. Max Tokens: This determines how much text the AI can work with in a single response. The default is 70 tokens, but feel free to adjust if you want longer responses.

6. Temperature: The temperature affects how random or controlled the AI’s answers will be.

• Lower values (e.g., 0.2) give predictable responses.
• Higher values (e.g., 0.8) let the bot become more creative.

“Train AI” Section

The Train AI Section, split into three main tools: Prompt Setting, Fine Tune, and Training List. Each part adds unique layers to the bot’s intelligence, shaping it for your needs.

1. Prompt Setting: In the prompt setting section, you will get, “Give Instructions to Chatgpt”, “Give Data/Content about your business/service/products/website”, and “Select the source for ChatGPT prompt analysis”. Let’s check them out

Give Instructions to ChatGPT: Here’s where you outline what the AI should be, do, and how it behaves.

What to Do: Provide instructions that shape the AI’s role. Example: “Act as a cheerful guide for troubleshooting WordPress issues, offering clear, concise answers.” Or: “Be a professional assistant who explains plugin functionality with patience and detail.”

Make the description precise to ensure the AI behaves as expected.

Give Data/Content About Your Business/Service/Products/Website: Fill this area with knowledge about what your business offers. This information feeds the AI with context to answer user queries.

Examples to Add:

• Business overview: “We create tools for WordPress users.”
• Product info: “Darklup is a dark mode plugin that lets you toggle themes effortlessly.”
• Features: “Works seamlessly across all devices and browsers, with customizable colors.”
• Keep entries relevant and useful for the AI to reference.

Select the Source for ChatGPT Prompt Analysis

This setting helps define where the AI pulls its insights from. Pick what aligns with your needs:

• Post: Let the bot analyze blog content to provide more contextual answers.
• Page: Use static pages to teach the AI about key topics or services.
• WooCommerce Product: Enable AI to access product descriptions and details for answering customer questions effectively.

Save Settings: After filling out all fields, don’t forget to hit Save Settings in the top-right corner. This locks in your instructions and data.

Through the Prompt Setting section, you create a flexible AI assistant tuned to your specific business needs. Keep refining the inputs for better results!

2. Fine Tune

This is where customization meets depth. Teach the AI precisely how to behave using datasets you control.

Training Type: Two paths are available:

1. Manual Input: Directly type information and examples into the system.
2. JSONL Upload: If you’ve got a pre-prepped dataset, upload a JSONL file to speed up the training.

Naming Your Training: Give your dataset a distinct label so you know what’s inside later. Example: “EchoBot Knowledge Base.”

For Train Manually: Here, you will see Add Data Set section.

Break training content into structured chunks:

• Role/Instructions: Describe how the bot should act. Example: “You’re a WordPress expert, focusing on plugins and themes.”
• Customer Question/Message: Add sample queries users might ask. Example: “Why isn’t my dark mode plugin working?”
• AI Response: Show the bot what a good answer looks like. Example: “Please check if the plugin is enabled in your settings. Let me guide you step by step.”

Hit Save Training Data after completing the entries.

For JSONL Upload: Here, all you need is to add a JSONL file in the file upload section, and then save.

3. Training List

Everything you’ve taught the bot so far will show up here. This is where you manage, tweak, or expand datasets.

What You’ll See:

• Training Name: The label for your dataset.
• File ID: A unique identifier for each file.
• Model Name: The specific OpenAI version used.
• Model Status: Whether training is ongoing, paused, or complete.
• Created: A timestamp of when the dataset was added.
• Actions: Tools to modify, remove, or edit datasets.

How to Use the Train AI

Training your AI involves a few structured steps that might look simple but can be shaped in diverse ways to unlock powerful functionality. Let’s explore how to prepare the AI for its role.

Step 1: Define the AI’s Role with a Prompt: Start with a well-crafted instruction—a roadmap for how your AI should interact with users. Clear, detailed prompts are like setting the stage for a play where the AI knows its lines perfectly.

• Example Role Setup: “You are a customer support engineer, adept at guiding users in event marketing. Your role is to provide clear, friendly assistance on event planning, promotion, and audience engagement, answering questions with patience and precision. You bring expertise to every interaction, offering actionable advice on strategies, registration platforms, and boosting participation rates.”
• Pro Tip: Make your instructions specific but leave room for adaptability. Let the AI shine where its insights matter most.

Step 2: Provide Data and Content About Your Business: Your AI performs better when it knows your world. Feed it information—your products, services, values, whatever defines your business—and watch it respond with greater depth.

• Input Suggestions: Describe your core offerings. Are you selling plugins? Services? Write it down. Detail your standout features. Example: “Darklup Plugin: Allows dark mode toggling, reduces eye strain, fully customizable for websites.”
• Outline benefits: From accessibility boosts to user satisfaction, cover it all.
• Why It Matters: AI gets sharper when its toolbox is packed with knowledge that matches your brand voice.

Step 3: Choose a Prompt Analysis Source: Point the AI to where it should dig for extra context—your blog, landing pages, or WooCommerce details.

What to Specify:

• Posts: Use blogs for extended learning.
• Pages: Reference FAQs or About Us pages.
• WooCommerce: Let it answer product-specific inquiries directly.
• Outcome: Accurate, context-rich responses tailored to each user’s query.

Step 4: Test the AI on Your Live Website: Get hands-on. After setting up prompts and data, try it out.

Tips for Effective Training

• Precision Wins: Don’t be vague—details make the AI shine.
• Stay Fresh: Update its database to reflect your newest offerings.
• Test Smart: Regular checks ensure your setup is on point.