Knowledge Base

Understanding Narrative AI Models and Data Usage

Understanding Narrative AI Models and Data Usage


Narrative's AI models, known collectively as Rosetta AI, provide a robust suite of automation tools geared towards enhancing data collaboration and data management processes. This article provides a detailed review of how these AI models function and what data they have access to, ensuring transparency and informed usage for all customers.

Rosetta AI Functionality

  1. Optional Interactions:
    • All interactions with Rosetta AI on the Narrative platform are optional.
    • These interactions are clearly marked with a “Rosetta AI” button in the platform.
    • Users can opt out simply by not utilizing these AI features.
  2. Data Privacy and Access:
    • Rosetta AI does not have access to customer data within customer data planes.
    • The AI interacts solely with the Narrative Control Plane and any metadata customers have provided to it.
    • The Narrative Control Plane contains metadata about customer datasets, including:
      • Manually generated metadata like dataset descriptions, samples, and statistics.
      • Dataset schema definitions including field names, descriptions, types, and enumerated values.
  3. AI Capabilities:
    • Rosetta Stone Automated Dataset Mapping:
      • Maps and aligns diverse datasets automatically to the Narrative Rosetta Stone attribute catalog.
    • Rosetta Automated Taxonomy Building:
      • Helps in creating and organizing taxonomies based on provided datasets.
    • Other Automations:
      • Various other AI-driven automation tasks to streamline data processes.

Model Hosting and Customization

  • Locally Hosted Models:
    • Rosetta AI capabilities are supported by a custom generative model hosted locally on the Narrative platform.
    • This model does not interact with external third-party APIs, ensuring that no data is sent outside the Narrative Control Plane.
  • Self-hosted Custom Models:
    • Customers have the flexibility to utilize self-hosted AI models.
    • These custom models can be fine-tuned for specific use cases as per customer requirements.

Rosetta AI Digital Assistant

  • Digital Assistant:
    • Customers have access to Rosetta AI’s Digital Assistant, which provides an AI-driven chat interface.
    • This functionality utilizes external AI APIs (e.g., OpenAI) and potentially other future providers.
    • Interaction with the Digital Assistant is entirely optional and can be opted out of by choosing not to interact with it.

In summary, while Rosetta AI provides powerful automation tools on the Narrative platform, all interactions are optional and clearly marked for user discretion. The AI models strictly operate within the boundaries of the Narrative Control Plane, respecting customer data privacy and offering customizable solutions.

< Back

Hi! I’m Rosetta, your big data assistant. Ask me anything! If you want to talk to one of our wonderful human team members, let me know! I can schedule a call for you.