Perplexity AI

Perplexity AI is an American artificial intelligence company that develops an AI-powered answer engine — a web search and synthesis product that processes user queries using large language models grounded in real-time web content and returns concise, cited responses. Founded in August 2022, the company grew rapidly to become one of the most widely used consumer AI products globally, reaching 22 million active users and processing 780 million queries per month by September 2025, at which point it was valued at USD 20 billion.^[1]

Company History and Founding

Perplexity was co-founded by Aravind Srinivas (CEO), Denis Yarats, Johnny Ho, and Andy Konwinski. Srinivas previously worked at OpenAI and Google DeepMind. The company was built on the premise that the traditional ten-blue-links search interface was fundamentally mismatched with how people actually want to retrieve information — through direct, synthesised answers rather than lists of documents to read separately. The company raised successive funding rounds at accelerating pace: a USD 73.6 million Series B in 2023, a USD 520 million Series D in June 2024, and a USD 500 million raise at a USD 9 billion valuation in November 2024. By September 2025 its valuation reached USD 20 billion, reflecting the rapid growth of AI-powered search as a category.

Core Product: The Answer Engine

The Perplexity answer engine accepts natural language queries and responds with a synthesised answer drawn from live web search results, with inline citations linking to source documents. Unlike a standard web search, which requires the user to evaluate and read multiple pages, Perplexity attempts to distil the relevant information directly. Users can follow up with additional questions in a conversational thread, with the system maintaining context. The product offers different modes: Pro Search applies more aggressive research strategies with multiple sub-queries and synthesis steps for complex questions; Quick Search prioritises speed for simpler queries. Pro subscribers gain access to a choice of underlying LLMs, including OpenAI's GPT series, Anthropic's Claude, Google's Gemini 2.5 Pro, and xAI's Grok, in addition to Perplexity's own Sonar models.

Sonar Model Family

In February 2025, Perplexity launched the Sonar model family — its in-house LLMs built on LLaMA 3.3 70B and optimised for high factual accuracy in web-grounded contexts.^[2] The family includes:

• Sonar: The base model, optimised for fast, accurate web-grounded responses.

• Sonar Pro: A more capable variant for complex, multi-step research queries.

• Sonar Reasoning: Powered by DeepSeek R1's reasoning architecture for analytical tasks.

• Sonar Deep Research: Designed for extended multi-step research synthesis.

The Sonar API, launched in 2025, provides developers access to this same search-grounded generation infrastructure, enabling third-party applications to build on Perplexity's real-time web retrieval stack.

Comet Browser

In July 2025, Perplexity launched Comet, a standalone web browser built on the Chromium platform with Perplexity's AI search as the default engine.^[3] Comet integrates agentic capabilities — performing multi-step web tasks on behalf of users — directly within the browser interface, positioning Perplexity in competition with both traditional browsers and emerging AI computer-use products.

Business Model

Perplexity operates a freemium model: a free tier provides access to the basic answer engine with the Sonar model, while the paid Pro tier (subscription) enables access to premium models, longer context, file uploads, and higher query limits. The Enterprise Pro tier offers API access, SSO, and data privacy guarantees for organisations. The Sonar API is priced on a per-query basis for developers.

Competitive Position

Perplexity competes with Google Search (which launched AI Overviews in 2024), Microsoft Bing with Copilot, and dedicated AI research tools such as Elicit and Consensus. Its differentiation lies in the quality of source citation, the breadth of model choice for Pro users, and the speed of information synthesis for information-dense queries. Perplexity AI has attracted a growing user base in Malaysia as professionals, students, and researchers adopt AI-powered search as a supplement or alternative to traditional search engines. Its ability to synthesise information from multiple sources with citations aligns well with the needs of Malaysian knowledge workers who frequently need to navigate English, Malay, and Chinese-language sources. In Malaysian universities, students at institutions such as Universiti Malaya (UM), Sunway University, and Taylor's University have begun using Perplexity for research tasks — a development that has prompted discussions among academic staff about AI literacy, citation practices, and the appropriate use of AI tools in academic work. The Malaysian Qualifications Agency (MQA) and individual institutions are developing guidance on this topic. Malaysian professionals in finance, law, and consulting sectors use Perplexity Pro for rapid research synthesis. The tool's strength in grounding answers in real-time web content is particularly valued in fast-moving domains such as financial markets, regulatory updates, and technology procurement research. Legal professionals at Malaysian law firms have noted its utility for preliminary case law research, though counsel remains cautious about verification given the possibility of AI-generated inaccuracies. For Malaysian AI developers and startups, the Sonar API offers an accessible entry point to building web-grounded AI applications. MDEC's ecosystem of Malaysia Digital-status companies includes AI startups exploring Perplexity's API as an alternative to building their own retrieval stack from scratch. The availability of the API through AWS Marketplace further reduces friction for companies already operating on AWS cloud infrastructure. From a regulatory perspective, Malaysia's Personal Data Protection Act (PDPA) applies to any AI service that processes Malaysian user data. Businesses deploying Perplexity-based solutions for customer-facing applications must ensure appropriate data handling agreements are in place — a consideration that the Personal Data Protection Department (JPDP) has flagged in its guidance on third-party AI tool procurement.

1. Built In. (2025). What Is Perplexity AI? Inside the AI company remaking internet search. Built In.
2. Perplexity AI. (2025). Introducing the Sonar model family. Perplexity AI Blog.
3. TechCrunch. (2025). Perplexity launches Comet, a Chromium-based browser with built-in AI search. TechCrunch.
4. Contrary Research. (2025). Perplexity Business Breakdown & Founding Story. Contrary Research.