Shatter the Veil

Organize Your Work with Sessions

Our platform allows you to organize your work into sessions, enabling you to manage multiple trains of thought simultaneously. These sessions are saved automatically, allowing you to pick up right where you left off at any time. A typical use case is when you're focused on a particular topic and come across an unrelated discovery that you want to explore later. Simply create a new session, and you can return to your original work or your new discovery whenever you wish.

Explore Over 400 Categories with Ease

We offer a wide range of over 400 categories to explore. To make navigation easier, our category search feature lets you quickly find category names without the need to scroll through the entire list. For example, if you're interested in the topic Egypt, you can search for it and instantly see all categories that include Egypt as a subcategory. This might include categories like Ancient History/Ancient Civilizations/Egypt and Coast to Coast/Ancient Civilizations/Egypt, among others.

Building Your Context

After selecting your categories, the next step is to build your context. The context is a carefully compiled collection of your search results that serves as a knowledgebase for the AI-LLM. It's essential to take the time to construct this context according to your specific needs and requirements, as it directly influences the quality of insights you'll gain.

You can build your context by searching the public knowledgebase using either semantic search or a Google-like full-text search. Once you've gathered your search results, you'll have the ability to filter, select the most relevant results, and even edit them to ensure they meet your standards.

Perform semantic search (natural language) or Google like full text search on the artifacts in your selected categories.

Then ask the AI-LLM to summarize the search results or answer your questions to gain deeper insights.

Semantic Search

You have the ability to control the relevancy of your search results by adjusting the Similarity Score. A higher score will yield more relevant results, ensuring that the search hits closely match your criteria. Additionally, you can limit the number of results by setting the Top K value. If you wish to retrieve all available hits, simply set the Top K value to 0.

Full Text Search

You can customize your searches by combining, excluding, or specifying the proximity of terms to refine the relevance and accuracy of your results.

Managing Your Context and Token Limit

As you generate search results, you can selectively add the most relevant ones to your Context. This Context will form the knowledgebase that the AI-LLM uses to respond to queries. The Context is subject to a size limit, known as the context window, which is measured in tokens. A token is roughly equivalent to 3/4 of a word, and the Research Desk will automatically update your token count as you add or remove results from your Context.

To help you stay within your token limit, you can make your results more concise. This process preserves the original meaning while reducing word count, allowing you to include more content without exceeding your token limit.

Editing Your Results

You also have the ability to edit your search results. This feature allows you to refine the content according to your judgment—whether that means clarifying, correcting, or removing parts of the results. These edits are temporary and will not be saved back to the public knowledgebase; they are solely for the purpose of building your Context for querying the AI-LLM.

Summarizing Your Context

Once you've built your Context, you have the option to summarize all the results within it. Summarizing can help you stay within your token count, making your query to the AI-LLM more efficient. However, if you feel that summarizing might lead to a loss of important meaning, you can choose to submit your Context as-is.

The Stash: A Workspace for Saving Results

The Stash is a workspace where you can store results that you find interesting but don't want to include in your Context right away. It allows you to save these results for later use. You can easily move results from the Stash to your Context and vice versa as your needs change.

Interacting with Multiple AI-LLMs

After building your Context, you can begin interacting with multiple AI-LLMs. This process is similar to working with ChatGPT, but with a key difference: The Research Desk submits your Context as a knowledgebase for the AI-LLM to process. You can choose to have the AI-LLM base its reasoning solely on this provided knowledgebase or to use it alongside its own training data.

Additionally, you have the option to control the AI-LLM's behavior by adjusting the temperature setting, which influences how creative or conservative its responses will be.