As search engine incumbents — namely Google — augment their platforms with AI technology, startups are trying to reinvent AI-powered search from the ground up. Could look like a work of Sisyphus, going up against competitors with billions billions of users. But this new breed of search startups believe they can carve out a niche, however small, in providing a superior experience.
One of the cohort, Perplexity AIthis morning announced that it has raised $70 million in a funding round led by IVP with additional investment from NEA, Databricks Ventures, former Twitter VP Elad Gil, Shopify CEO Tobi Lutke, former GitHub CEO Nat Friedman and Vercel founder Guillermo Rauch. Other participants in the round included Nvidia and — notably — Jeff Bezos.
Sources familiar with the matter tell TechCrunch that the round values Perplexity at $520 million post-money. This is a big change in Kingdom of AI startups. But considering Perplexity has only been out since August 2022, it’s an impressive climb nonetheless.
Perplexity was founded by Aravind Srinivas, Denis Yarats, Johnny Ho and Andy Konwinski — engineers with backgrounds in artificial intelligence, distributed systems, search engines and databases. Srinivas, CEO of Perplexity, previously worked at OpenAI, where he researched language and artificial intelligence models along the lines of Stable Diffusion and DALL-E 3.
Unlike traditional search engines, Perplexity offers a chatbot-like interface that allows users to ask questions in natural language (eg, “Do we burn calories while we sleep?”, “What is the least visited country? ” and so on). The platform’s AI responds with a summary containing source citations (mostly websites and articles), so users can ask follow-up questions to dig deeper into a particular topic.
“With Perplexity, users can get instant … answers to any question with full sources and references,” Srinivas said. “Embarrassment concerns anyone and everyone who uses technology to seek information.”
The Perplexity platform is based on a series of next-generation AI models developed in-house and by third parties. Subscribers to Perplexity’s Pro plan ($20 a month) can switch models — Google’s Gemini, Mistra 7Bl, Anthropic’s Claude 2.1, and OpenAI’s GPT-4 are currently in rotation — and unlock features like the creation of images. unlimited use of Perplexity’s Copilot, which takes personal preferences into account during searches. and file uploads, which allows users to upload documents including images and have models analyze the documents to formulate answers about them (eg “Summary of pages 2 and 4”).
If the experience sounds comparable to Google’s Bard, Microsoft’s Copilot, and ChatGPT, you’re not wrong. Even Perplexity’s chat-promoting user interface is reminiscent of the most popular AI tools of our time.
Beyond the obvious competitors, search engine startup You.com offers similar AI-powered source summarization and reporting tools, optionally with GPT-4.
Srinivas argues that Perplexity offers more powerful filtering and search discovery options than most, for example, allowing users to narrow searches to academic papers or browse popular search topics submitted by other users on the platform. I’m not convinced it is So differentiated that they could not be reproduced — or not already reproduced for that matter. But Perspective has ambitions beyond search. It is starting to serve its own AI models, which leverage Perplexity’s search index and the public web for seemingly improved performance, through an API available to business customers.
This reporter is skeptical about the longevity of next-generation AI search tools for a number of reasons, not least of which is that AI models are expensive to run. At one point, OpenAI was spending approximately $700,000 per day to keep up with demand for ChatGPT. Microsoft is said to be losing Meanwhile, an average of $20 per user per month on the AI code generator.
Sources familiar with the matter tell TechCrunch Perplexity’s annual recurring revenue is currently between $5 million and $10 million. That seems pretty healthy… until you consider it million dollars It often costs to train generational AI models like Perplexity’s.
Concerns about misuse and misinformation inevitably arise around next-gen AI search tools like Perplexity, too — as they should. AI isn’t the best summarizer after all, sometimes key details are missingmisinterpreting and exaggerating language or otherwise inventing facts very authoritatively. And it’s prone to spewing bias and toxicity—like Perplexity’s own models recently proved.
Another potential speed bump on Perplexity’s road to success is copyright. Generation AI models “learn” from examples to build essays, codes, emails, articles, and more, and many vendors — including Perplexity, presumably — are scraping the web for millions to billions of these examples to add in the training datasets. Vendors argue that the fair use doctrine provides a blanket protection for web-scraping practices, but artists, writers and other copyright holders disagree — and have filed lawsuits seeking damages.
As a tangential aside, while a growing number of AI generation vendors offer policies that protect customers from IP claims against them, Perplexity does not. According to the company data Terms of usecustomers agree to “hold harmless” Perplexity from claims, damages and liabilities arising out of the use of its services — meaning Perplexity is off the hook for legal fees.
Some plaintiffs, such as the New York Times, have argued that the generation’s AI search experiences reduce publishers’ content, readers and ad revenue through anticompetitive means. “Anti-competitive” or not, technology is definitely affecting traffic. A model from The Atlantic were found that if a search engine like Google incorporated artificial intelligence into search, it would answer a user’s query 75% of the time without requiring a click to their website. (Some vendors, such as OpenAI, have signed deals with some news publishers, but most — including Perplexity — have not.
Srinivas presents this as a feature – not a bug.
“[With Perplexity, there’s] you don’t have to click on different links, compare answers or search endlessly for information,” he said. “The era of SEO spam, sponsored links and multiple sources will be replaced by a more efficient model of knowledge acquisition and sharing, pushing society into a new era of rapid learning and research.”
The many uncertainties surrounding Perplexity’s business model — and AI and consumer search in general — don’t seem to be deterring its investors. To date, the startup, which claims ten million active monthly users, has raised more than $100 million — much of which is being used to expand its 39-person team and build new product features, Srinivas says.
“Perplexity is strongly building a product capable of bringing the power of artificial intelligence to billions,” Cack Wilhelm, general partner at IVP, added via email. “Aravind has the unique ability to maintain a big, long-term vision while shipping product relentlessly, demanding to address a problem as important and fundamental as search.”