Privata Docs

Appearance

Building Your Knowledge Base ​

The Knowledge Base is your company's library of documents and websites that Privata uses to provide context to its AI agents and tools. This context helps the platform generate better answers, proposals, summaries, and images such as architecture diagrams.

The Knowledge Base is organized into collections. Each collection groups related material together so the right information can be used for the right activity. Your Knowledge Base also has an Approved Answers tab, where your best vetted responses are stored for reuse. See Approved Answers.

INFO

Collections belong to your company workspace, not to an individual user. Any member of your company can use the collections your team has built, which keeps supporting material consistent across projects.

What a Collection Is ​

A collection is a specific grouping of documents and website content inside the Knowledge Base, used to organize supporting material that can later be associated with a project.

Good collection examples include:

  • A specific product and its user guides
  • Company background information and capability statements
  • Pricing tables
  • Completed RFxs and past performance
  • Statements of work
  • Technical reference documents

You do not need a separate collection for every product or topic. Privata retrieves the most relevant passages from across your collections and uses labels to focus results, so a few broad collections are usually easier to manage and work just as well. Organize around major areas (such as company knowledge, product and pricing material, and past performance) and let labeling do the fine-grained sorting.

TIP

Keep collections broad and lean on labels and your Company Profile Settings rather than splitting content into many small collections. When you work on a project, attach the collections it needs, often several or all of them, or set its Retrieval scope to All collections.

How Collections Work with Projects ​

When you work on a project, you select the collections that should be used as context and set the project's Retrieval scope: either the collections listed on the project, or all collections in your company Knowledge Base. This lets Privata use only the most relevant material for that response effort.

  • Projects hold your source documents, such as RFxs.
  • Collections hold the supporting Knowledge Base content used as context.
  • The selected collections inform the AI agents and tools when generating answers, proposals, and other project content.

Creating a Collection ​

When creating a collection, a name is required. You can also add an optional description, which is recommended so users can quickly understand what content is included.

Collections can be created from several places:

  • From the Knowledge Base page in the top navigation, using Create New Collection
  • From an existing project using Add Collection
  • From the New Project intake when you set up a project

When you create a collection you will see a short "What happens next" explainer: you upload documents, Privata processes them with AI, and the content becomes available for semantic search and RFx analysis.

Uploading and Indexing Documents ​

When a document is uploaded to a collection, Privata indexes it. Indexing breaks the document into smaller sections (words, phrases, sentences, and content groupings) that can be searched and matched by meaning. This lets the platform retrieve the most relevant parts of your material and use them as supporting context during AI generation.

While a document indexes, Privata also applies labels to it (see below). Upload progress and labeling status are shown as the document is processed, and the upload dialog closes as soon as your files are queued so you can keep working.

Duplicate files ​

If you upload a file with the same name as one already in the collection, Privata detects the duplicate and asks whether you want to overwrite the existing document or keep both. The same check runs when you add documents from the project page.

Labels ​

Labels are a controlled vocabulary that describes what each document and collection is about: for example the product or service it covers, the customer sector, the compliance framework, or the document type. Labels sharpen retrieval so the AI pulls the most relevant content for a given project.

  • Privata can auto-label documents as they are indexed; you can turn this on or off.
  • Use Edit Labels to adjust the labels on a collection or document.
  • Labels refresh live while a document is being indexed.
  • Your company's label vocabulary can be curated over time, and unused values can be retired.

Project labels (Response Style, Products/Services & Offerings, Customer Sector, Certifications) work together with document labels to focus retrieval and set the response style for a project. The available values come from your Company Profile Settings.

Supported File Types ​

Privata supports the following for Knowledge Base content:

  • PDF
  • Word (DOC and DOCX)
  • Excel (XLS and XLSX) and CSV
  • PowerPoint (PPT and PPTX)
  • XML and Markdown (MD)
  • Images (JPG, JPEG, PNG): text is extracted with OCR
  • HTML and website URLs

Additional document types continue to be added over time. If you need support for a file type that is not listed, contact support@sky-path.ai.

Adding Website Content ​

Collections can also include website content. Use the Add Website button and enter the URL of the page you want to include. Privata indexes the content found at that link so it can be used as part of your Knowledge Base. This is useful for:

  • White papers
  • User guides
  • Case studies
  • Public reference material

Bringing in Content from Connected Services ​

You can import files into your collections from connected services (Google Drive, SharePoint, and Salesforce) without permanently storing the originals. Connect these services from your Account Settings, then choose the files you want to bring in.

Searching Your Knowledge Base ​

Use Knowledge Base Search to find content across your collections. Results are ranked by relevance and paginated, you can filter by collection, and you can Preview a result before opening it. See Knowledge Base Search.

Best Practices for Managing Collections ​

  • Organize a few broad collections around major subject areas, and rely on labels rather than splitting content into many small collections.
  • Add descriptions so other users understand the purpose of each collection.
  • Attach only the collections that are relevant to the current project.
  • Keep product, pricing, background, and past-performance material grouped in a way that matches your workflow.
  • Review collection contents regularly so outdated information does not affect AI output.

Deleting a Collection ​

DANGER

When you delete a collection, all indexed data in that collection and all associated uploaded documents are permanently removed. This action cannot be undone.

What to Do Next ​

After creating a collection, the next step is usually to attach it to a project and begin using it with your source documents and AI tools.

For related guidance, continue to the Working with Projects and New Project pages.