'Future of Search in the Age of AI' Perplexity CEO, Aravind Srinivas | RAISE Summit 2024 | Paris

Company Positioning & Vision

• Perplexity AI is defined as an "answer engine" rather than a traditional search engine or general chatbot, designed to provide tailored, source-cited answers to specific questions rather than lists of links.
• CEO Aravind Srinivas established the company based on a "data flywheel" concept where product usage generates data to further improve AI accuracy.
• The founding team was inspired by DeepMind's 2019 environment and the philosophy that AI progress and product quality should be inextricably linked.
• The core team composition includes researchers (Srinivas and CTO Denis Yarouts) paired with a "hardcore" competitive programming expert (Johnny Ho) and an operational veteran (Andy, co-founder of Databricks) to ensure technical, product, and scaling maturity.

Market Strategy & Competitive Advantage

• Perplexity explicitly targets the gap in current search (Google) where the primary business model (advertising) creates a misalignment between shareholder interests and user utility.
• The strategy leverages Jeff Bezos' "margin is opportunity" principle, exploiting the fact that incumbents like Google are disincentivized to disrupt their high-margin ad business with lower-margin, higher-cost AI answer products.
• Unlike general-purpose models (ChatGPT, Gemini), Perplexity focuses exclusively on the search use case to minimize "hallucinations," treating them as a critical bug rather than a creative feature.
• The product differentiates itself by strictly adhering to cited sources from the web, allowing users to verify information, thereby prioritizing accuracy and neutrality over open-ended creativity.

Business Model & Monetization

• Current operations involve lower margins per query compared to traditional search due to higher infrastructure costs, with no established advertising model for the answer-engine format.
• The long-term goal is to emulate the Google/Yahoo shift by creating a new advertising or consumption model that aligns shareholder value with user correctness, prioritizing user alignment before monetization.
• Perplexity intends to maintain neutrality by aggregating diverse sources and high-authority content, presenting multiple viewpoints on controversial topics rather than generating AI opinions.

Technology & Model Agnosticism

• The platform employs a routing and orchestration system to select the appropriate AI model for specific query types (e.g., using APIs for weather vs. large models for tax questions) rather than relying on a single monolithic model.
• Aravind Srinivas argues that the AI model layer is becoming commoditized, with average users unable to distinguish performance differences between top-tier models like GPT-4 and Opus.
• While the default experience automatically selects models, the platform offers user choice for advanced users, though the company does not intend to push specific partners for business reasons.

Future Outlook & Narrative Arc

• The company plans to continue refining the "answer engine" format, potentially introducing new advertising mechanisms once the product form factor is established.
• Srinivas views the current state of search as a transition from "10 blue links" (1990s) to direct answers (2020s), positioning Perplexity as the definitive solution for this new phase.
• The founders intend to scale the company without over-hiring, relying on the "adult in the room" guidance to maintain process discipline during rapid growth.