Getting started with AI can be overwhelming. We’ll help you understand the jargon and basic tips for using WDChat.
You can also check out our feature-specific tutorials and find additional information in our full FAQ.
WDChat basics
How can I use WDChat?
How do I access past chats?
Browse past conversations using the chat history panel on the left.
You can pick up a conversation where you left off and WDChat will remember what you were last talking about.
How do I rename chats?
To rename old chats, select the three dots next to the chat name > Rename.
Enter a new name and select Save.
How do I delete chats?
To delete old chats, select the three dots next to the chat name > Delete > Delete.
Note: This will only delete the conversation from your Chat History panel. Western Digital stores all user interactions with WDChat.
Can I make images or videos, upload files, or have it analyze a URL?
When toggled to Public internet data, you can upload certain types of text or image files with up to highly confidential Western Digital data in WDChat. There is a 1GB limit for uploads. We’re working on supporting other file types soon. Check our roadmap for more details.
Can I ask questions about Western Digital?
Yes. When you enter a prompt that requires Western Digital data, WDChat’s answers will utilize information from WesternDigital.com, IT and P&C knowledge articles from HelpZone, Apps, Western Digital news and events
Can I use Western Digital data in my prompts?
WDChat is approved for data up to Highly Confidential.
Learn more about what AI tools you can use on your Western Digital device.
How do I turn on Dark Mode?
Select your name > Appearance > and your preferred theme.
Where can I find more information?
Visit our full FAQ page or contact the WDChat Support Teams space.
How do I write a good prompt?
Check our prompt writing guide for tips.
What is an LLM and how does it work?
What is an LLM?
LLM stands for Large Language Model. They are advanced AI systems designed to process and generate human-like text based on the input they receive.
How are LLMs trained?
LLMs are pretrained by feeding the model large datasets to help it learn how to predict the next word in a sentence. This also helps it learn how to interpret user queries.
They are then fine-tuned on specific tasks or additional datasets to make the base model more useful for the desired use case.
LLMs don’t think so much as they predict. They sound so convincing because they’ve been trained on enormous amounts of natural language.
What is Microsoft Azure OpenAI?
A cloud-based Microsoft service that gives businesses access to OpenAI’s AI models without having to train and deploy them entirely on their own.
What does GPT stand for?
Generative: The ability to generate new content.
Pre-trained: It’s trained on large amounts of data to learn language rules and context before it’s asked to perform specific tasks.
Transformer: A term for the underlying machine learning model, the way it’s been programmed.
Which models does WDChat use?
The full version of WDChat uses several models, including GPT-4o, GPT-o4 mini, GPT-4.1 mini, GPT-5, GPT-5 mini, GPT-5 chat, GPT-5 nano, Gemini 2.5 Flash, Gemini 2.5 Pro, and Gemini 2.0 Flash with live web data.
WDChat will detect which model to use based on your prompt.
How up to date is the information?
Gemini Flash provides current information via live web data in WDChat. Based on your prompt, WDchat will determine the best model to respond with.
WDChat provides up-to-date, published information from Cisco data sources.
What is a prompt?
A prompt is the text you enter to start the chat. It could be a simple question or a complex request.
Check our prompt writing guide for tips.
Is it always accurate?
No. LLMs are still learning and they make mistakes. Sometimes these mistakes are due to outdated training data or data with errors or biases.
Remember, LLMs are predicting words in a sentence, they are not thinking.
It’s always good to check WDChat’s answers.
What is a hallucination?
Hallucinations are inaccurate, irrelevant, often nonsensical responses. Bias or gaps in the training data or misinterpretation of prompts may lead to hallucinations.
Remember, LLMs are predicting words in a sentence, they are not thinking.
It’s always good to check WDChat’s answers.
What is a token?
A token is the basic unit of data used by an LLM. Tokenization is the process of breaking down text-based prompts into individual tokens. This is how LLMs learn to interpret and generate text.
Is one token equal to one word?
Not exactly. 1 token is roughly equivalent to 4 characters, but this can vary.
One token may be equal to one word, a part of a word, or an individual character.
What is WDChat's token limit?
If you’re using WDChat in the browser, there is an 8,000 token limit for prompts, which is roughly 36,000 characters.
If you’re using WDChat in Teams, Webex’s 5,400 character maximum will apply.