The Vector Vault is a centralized knowledge repository within Prompt Privacy that enables organizations to store, organize, and interact with their data in an intelligent and searchable format. It transforms raw documents and datasets into vectorized representations, allowing users to perform semantic search and retrieve contextually relevant insights with high precision. Instead of relying on exact keyword matching, Vector Vault understands meaning and relationships within data, making it possible to surface relevant results even when queries are conceptually similar rather than identical.
This feature is designed for managing unstructured and semi-structured content such as documents, PDFs, text files, and other knowledge assets. By grouping data into dedicated vaults, users can structure information around specific projects, teams, or workflows while also enabling secure collaboration through controlled sharing. Vector Vault empowers users to quickly access insights through natural language queries, uncover hidden patterns, and enhance decision-making with data-driven intelligence.
Vector Vaults store and organize information by converting uploaded content into vectorized data representations, enabling advanced semantic search and similarity-based retrieval. Each vault acts as an independent repository where files, documents, and datasets are indexed as vectors, allowing the system to interpret meaning, context, and relationships across data points.
This makes Vector Vault suitable for handling structured, semi-structured, and unstructured content. Whether working with spreadsheets, mixed-format business files, or documents such as PDFs, text files, and reports, users can organize and query all data types within a unified system. Vaults can be created for specific projects or workflows, shared securely with team members, and accessed using natural language queries to retrieve meaningful insights quickly and efficiently.
Once files are uploaded into a vault, users can interact with them directly through the chat interface. You can select either an entire vault or specific files within a vault, depending on whether you want a broad or focused analysis.
When a prompt is submitted, the system performs a semantic search across all files within the selected vault. It identifies the most relevant content based on contextual similarity and uses the best-matching file to generate a response. To ensure accuracy and clarity, responses are grounded in a single most relevant file at a time.
Users can ask questions about concepts, themes, or specific details present in their documents. As long as the vault is selected, the system dynamically interprets queries and retrieves contextually relevant information across the stored data. This enables continuous exploration of multiple ideas within the same knowledge space.
For best results, prompts should be clear, concise, and contextually descriptive, ensuring the model can accurately retrieve the most relevant insights.
Vector Vault supports a wide range of file formats, enabling seamless ingestion of diverse content types. Supported formats include: PDF, TXT, DOC, DOCX, XLSX, RTF, JSON, CSV, MD, PPTX, and PPT.
<aside>
Note on CSV Files: CSV files are not ideally suited for Vector Vaults due to their rigid tabular structure, which limits semantic vectorization. For best results, users working with CSV data should either leverage the Cognitive Storage Engine (CSE) or copy relevant sections of the data directly into the chat interface. This ensures the system can process and interpret the content more effectively, especially for analytical or contextual queries.
</aside>