Skip to main content

Vito AI Assistant

Introduction​

Vito AI Assistant is an embeddable, multilingual chatbot designed to enhance the reading and learning experience for users interacting with digital publications on the Publica.la platform. Leveraging advanced AI capabilities, Vito helps readers understand, summarize, and interact with complex content through a conversational interface seamlessly integrated inside publication readers as an iframe.

Vito empowers readers to:

  • Instantly generate summaries of what they are reading
  • Create self-assessment quizzes from book content
  • Ask free-form questions about the current section
  • Interact in English, Spanish, or Portuguese, with the assistant adapting automatically

This assistant supports educational, academic, and professional use cases, providing support for students, researchers, educators, and lifelong learners.

Key Features​

1. Contextual Summarization​

Vito enables users to obtain concise, contextual summaries of publication content on-demand:

  • Instant Summaries: With a single click ("Summarize page"), users can generate AI-curated summaries of the current page or pages.
  • Custom Summaries: Users may select start and end pages and choose between summary formats, such as paragraph or bullet-point key points.
  • Expanded Summaries: After receiving a standard summary, users can opt for a more detailed (expanded) summary for deeper understanding.

Supported Summary Types and Limits​

  • Paragraph: Cohesive, well-structured abstract suitable for quick assimilation.
  • Key Points (Bullet List): Concise list of main ideas or facts.
  • Page Range: Summaries can span up to 10 pages at a time for custom summaries.

2. Self-Assessment Quizzes​

Vito empowers users to test their comprehension using AI-generated assessments:

  • Configurable Assessments: Readers can choose the page range and number of questions (3 or 5, up to a 10-page range).
  • Multiple-Choice Questions: Each question includes three options (A, B, C), with only one correct answer and plausible distractors.
  • Immediate Feedback: Users receive instant scoring, as well as concise justifications for correct answers to reinforce learning.
  • Result Review: Upon completion, Vito presents a score and detailed review of each question with explanations.

Assessment Content​

  • Assessments are based strictly on the selected pages' text, ensuring questions are relevant and tailored to what was read.

3. Organic Questions (Free-Form Q&A)​

Vito enables users to ask any question about the content they are reading—in natural language:

  • On-Page Context: The assistant answers based only on the section currently being read (the context window).
  • Smart Context Range: Vito considers a window of approximately 4 pages before and 8 pages after the current page(s), with a maximum of 20 pages per request (a technical limitation).
  • Use Cases: Ask for definitions, explanations, clarifications, or additional insights as needed—with answers generated solely from the book's actual content.

Technical Limit​

  • Context window cap: When answering organic questions, the system never reviews more than 20 pages at a time, even if a broader range is specified. This ensures prompt responses and optimal performance.

4. Multilingual Support​

  • Languages Supported: English, Spanish, and Portuguese.
  • Language Selection: Vito starts in the user's device language but allows users to switch languages at any time using the language switcher in the chat header.
  • Session Persistence: Language preference is remembered for the entire reading session and saved locally for future sessions.
  • Consistent Communication: All menus, prompts, and AI-generated content (summaries, answers, assessments) are delivered in the selected language, regardless of the book's original language.
  • Immediate Updates: Changing the language instantly updates all visible interface elements and affects all future AI responses.

5. Seamless Integration and UI​

  • Embedded Experience: Vito appears as a compact chat panel next to the reading interface, maintaining an unobtrusive presence.
  • Menu-Driven and Conversational Flows: Key actions are accessible via buttons, while the chat allows for free-form interaction.
  • Language Switcher: A globe icon in the chat header provides quick access to language selection (English, Spanish, Portuguese) with instant UI updates.
  • Theme Synchronization: Vito automatically matches Volpe's theme (light/dark) for a consistent visual experience, with real-time updates when the theme changes.
  • Conversation Persistence: (In development) Vito will remember past sessions per user and publication, allowing users to resume their learning journey seamlessly.

Privacy, Security, and Content Rights​

Opt-In AI Functionality​

Vito AI Assistant represents a strictly opt-in feature that is disabled by default. We do not use AI to automatically process content without explicit user consent. The AI functionality is designed to enhance personal interaction with content rather than extract, replicate, or create derivative works from publications.

Data Processing and Privacy​

Our AI implementation follows strict privacy and security protocols:

  • Ephemeral Processing: All AI interactions are processed in real-time within the user's session scope and are not stored or reused beyond the session.
  • Limited Data Transmission: Only the user's specific selection and very limited metadata are sent to the AI model—never the entire publication content.
  • Encrypted Communication: All content travels encrypted back and forth between the user's device and our servers.
  • No Third-Party Storage: Content is not stored by any third-party providers during AI processing.

Content Rights and AI Training​

We maintain strict boundaries regarding content rights and AI model training:

  • No Training Data: We do not use any book content to train AI models. Our AI provider, OpenAI, is specifically configured not to retain or use any data from our platform for training purposes.
  • Prompt-Based Interactions: All AI-powered features rely solely on prompt-based interactions rather than content ingestion or analysis.
  • No Derivative Works: We do not read, copy, or create derivative works of content using AI. The assistant only responds to direct user queries about content they are actively reading.
  • Content Integrity: AI features are designed to enhance personal understanding and engagement without compromising the original work's integrity or rights.

Current Implementation Status​

Vito is currently in beta release with controlled rollout. During this phase, the functionality is enabled for all content within participating tenants. Our technical and functional design prioritizes content protection and user privacy, with additional controls available for implementation on a case-by-case basis as needed.

User Workflow Examples​

Changing Language​

  1. Click the globe icon (🌐) in the chat header.
  2. Select your preferred language: English, Spanish, or Portuguese.
  3. The interface immediately updates and all future AI responses will be in the selected language.

Summarizing Content​

  1. Click "⚡️ Summarize page" to receive an instant summary.
  2. Opt for "Custom summary" to specify a range and summary style.
  3. Request "Expand summary" for more depth if required.

Self-Assessment​

  1. Select "🧠 Self-assessment."
  2. Configure the desired pages and question count.
  3. Answer questions, review your score and explanations.

Asking Organic Questions​

  1. Click "💬 Ask a question."
  2. Type your question about the section.
  3. Receive a focused, contextual answer drawn from up to 20 pages of current and nearby content.

Important Technical Notes​

  • Online only: Vito currently requires an active internet connection.
  • Content type: Vito currently only works on PDFs.
  • Page Range for Organic Questions: Answers are limited to a 20-page context window, selected intelligently around the current reading position.
  • Page Range for Custom Summaries and Assessments: The user may specify up to 10 pages.

Frequently Asked Questions​

Does Vito access or expose content outside the publication?​

No. All answers are generated solely from the text of the publication, in accordance with territorial and licensing restrictions.

Does Vito store user data or conversations?​

Session persistence is in development. When enabled, conversations are securely stored per user and publication for resumption and analytics.

Is there a limit to how many questions or summaries a user can request?​

Yes. AI limitations and performance considerations (such as the 20-page context cap for organic questions and the 10-page cap for assessments and summaries) ensure a fast, meaningful experience.

Can Vito be embedded in any store or publication?​

Vito is available for integration with eligible plans on Publica.la and for tenants with compatible application versions. Contact support for enablement details.

Is Vito enabled by default for all users?​

No. Vito AI Assistant is a strictly opt-in feature that is disabled by default. Users must explicitly choose to enable and use the AI functionality.

Does Publica.la use AI to automatically process or analyze content?​

No, we do not use AI to process content automatically. We offer AI-powered functionality only through strictly opt-in features that enhance user engagement by allowing them to explain, expand, and summarize their own highlights, generate summaries of book sections, and perform self-assessments based on book content.

Is user data or content used to train AI models?​

No, we do not use any book content or user data to train AI models. All AI-powered features rely on prompt-based interactions, and our provider, OpenAI, is specifically configured not to retain or use any data from our platform for training purposes.

How is content protected during AI processing?​

Content protection is ensured through multiple safeguards: only user selections and limited metadata are transmitted (never full content), all data travels encrypted, processing is ephemeral with no storage beyond the session, and no third parties store the content during processing.

X

Graph View