Lecture on Perplexity AI and Its Applications

Introduction

• The founding belief was that foundational models will become commoditized, similar to storage or compute.
• Value seen in incredible search capabilities and integration with various data.
• Introduced the concept of an AI wrapper, which was a contrarian viewpoint a few years ago.

Use Cases of Perplexity

• Mainly used for finance verticals as everyday search.
• Replaces traditional search engines like Google by providing direct answers instead of multiple links.
• Breaks queries into sub-questions, searches in real-time, and ranks the best sources.
• Model agnostic; selects the best model to answer questions.

Example Use Cases

• News Inquiry: Handles questions about news events, provides political background, stance analysis, etc.
• Financial Data Integration: Integrates with third-party data sources like FactSet to analyze earnings and provide insights.
• Deep Research Functionality: Allows for complex query handling similar to a first-year analyst by using deep research mode.

Integration with Data Sources

• Connects with data sources like FactSet and Crunchbase for deep analysis.
• Users can authenticate subscriptions to access third-party data within Perplexity.

Features and Functionalities

• Fast Information Retrieval: Provides quick answers to queries, saving time compared to traditional search methods.
• Profile Pages and Dashboards: Offers features like live earnings transcripts, sentiment analysis, and financial dashboards similar to Google Finance.
• Spaces and Templates: Users can create spaces for repetitive tasks, save prompts, and share templates with teams.

Spaces Feature

• Customizable Search Spaces: Users can set pre-prompts, link to specific sources, and prioritize searches.
• Repetitive Task Automation: Allows automation and customization for tasks like building profiles or conducting analysis.
• Integration with Internal Documents: Users can upload documents manually or connect through Google Drive/SharePoint.

Privacy and Security

• Emphasis on data security with encrypted storage.
• Options to control data retention and permission settings to manage data access.

Model Selection and Prompt Engineering

• Users can choose different models for different depths of research.
• Perplexity aims to handle prompt engineering internally, allowing users to focus on query design.

Applications in Finance

• Useful for IPO valuation, peer comparison, and investment analysis.
• Provides detailed summaries and automates the analysis process.

Feedback and Future Developments

• Open to user feedback for integrating more data sources like Bloomberg or PitchBook.
• Continually updating and integrating new models and functionalities.