27. Perplexity
Founders: Aravind Srinivas (CEO), Denis Yarats, Johnny Ho, Andy KonwinskiLaunched: 2022Headquarters: San FranciscoFunding: $1.4 billion (PitchBook)Valuation: $9 billion (PitchBook)Key Technologies: Artificial intelligence, generative AI Industry: Enterprise technologyPrevious appearances on Disruptor 50 list: 0
Built by alumni from OpenAI, Meta, and Quora, Perplexity AI is attempting to create the next generation of search engines by combining generative AI with the internet.
In April, it expanded into new territory through a deal with Motorola, allowing it to widen its user base. Its technology will be included in Motorola's "Moto AI" capabilities. While not the first AI-powered search engine to partner with smartphones, it now is in direct competition with Apple and OpenAI's Siri-ChatGPT integration which was announced in December 2024.
″Search shouldn't be about endless links and ads — it should give the user directly what they want, and we think the best way to do that is through an answer engine," Perplexity CEO Aravind Srinivas said at a New York launch event. "Your phone is now an answer machine, personal assistant and a research agent."
Instead of pulling up links, Perplexity is essentially a hybrid between a chatbot and a search engine. It offers answers sourced directly from the Web, which it summarizes using large language models (LLMs). The platform's signature feature is its commitment to citation-backed responses, which not only provides context but factual backing. By the end of 2024, Perplexity was answering 20 million questions a day, according to the company.
The company's freemium model allows public access, and last year it added advertising. Brand partners like Indeed and Whole Foods joined the program, and it also launched a Publisher Program with partners like TIME, The Independent, and Fortune. It also has a paid tier that gives faster responses, PDF interpretation, and features like image generation. Its Enterprise Pro customers lean towards the finance and tech sphere, but it also works with other companies including CMA CGM, Nvidia, and the Cleveland Cavaliers. The company pulls in about $100 million in annual recurring revenue, sources told CNBC.
In May, it linked up with PayPal for an AI chat-based shopping feature that allows for booking travel, buying products and securing concert tickets on Perplexity's chat interface, paying instantly with PayPal or Venmo.
It has also made some bold moves this year, entering a bid in January to merge with TikTok under a company called NewCo. The deal would combine Perplexity and TikTok U.S., and would allow for the U.S. government to own up to 50 percent of the new company contingent upon a future IPO. In March, it laid out its TikTok vision in a blog post.
Because Perplexity depends on external sources it, like many web-based generative AI services, it can easily be swayed by the bias of the content it finds. Misinformation, or SEO-tailoring, can easily be taken as fact without additional checks. Despite this, Perplexity is gaining traction. It's backed by prominent investors including New Enterprise Associates, SoftBank Vision Fund, Jeff Bezos and Daniel Gross, a former Apple executive whose own search engine startup was acquired by the device giant.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
CNBC
25 minutes ago
- CNBC
German startup DeepL says latest Nvidia chips lets it translate the whole internet in just 18 days
DeepL on Wednesday said it was deploying one of the latest Nvidia systems that would allow the German startup to translate the whole internet in just 18 days. This is sharply down from 194 days previously. , DeepL is a startup that has developed its own AI models for and competes with Google Translate. Nvidia is meanwhile looking to expand the customer base for its chips — which are designed to power artificial intelligence applications — beyond hyperscalers such as Microsoft and Amazon. It also highlights how startups are using Nvidia's high-end products to build AI applications, which are viewed as the next step after foundational models, such as those designed by OpenAI. The Cologne-based company is deploying an Nvidia system known as DGX SuperPOD. Each of the DGX SuperPOD server racks contains 36 B200 Grace Blackwell Superchips, one of the company's latest products on the market. Nvidia's chips are required to train and run huge AI models, such as the ones designed by DeepL. "The idea is, of course, to provide a lot more computational power to our research scientists to build even more advanced models," Stefan Mesken, chief scientist at DeepL, told CNBC. Mesken said the upgraded infrastructure would help enhance current products like Clarify, which the company launched this year. Clarify is a tool that asks users questions to make sure context is incorporated in the translation. "It just wasn't technically feasible until recently with the advancements that we've made in our next-gen efforts. This has now became possible. So those are the kinds of advances that we continue to hunt for," Mesken said.
Yahoo
40 minutes ago
- Yahoo
Meta and TikTok challenge tech fees in second highest EU court
By Foo Yun Chee LUXEMBOURG (Reuters) -Meta Platforms and TikTok said a European Union supervisory fee levied on them was disproportionate and based on a flawed methodology as they took their fight with tech regulators to Europe's second highest court on Wednesday. Under the Digital Services Act that became law in 2022, the two companies and 16 others are subject to a supervisory fee amounting to 0.05% of their annual worldwide net income aimed at covering the European Commission's cost of monitoring their compliance with the law. The size of the annual fee is based on the number of average monthly active users for each company and whether the company posts a profit or loss in the preceding financial year. Meta told judges at the General Court it was not trying to avoid paying its fair share of the fee, but it questioned how the Commission had calculated the levy, saying it had been based on the revenue of the group rather than of the subsidiary. Meta's lawyer Assimakis Komninos told the panel of five judges the company still did not know how the fee was calculated. He said the provisions in the Digital Services Act, or DSA, "go against the letter and the spirit of the law, are totally untransparent with black boxes and have led to completely implausible and absurd results". ByteDance-owned Chinese online social media platform TikTok was equally critical. "What has happened here is anything but fair or proportionate. The fee has used inaccurate figures and discriminatory methods," TikTok lawyer Bill Batchelor told the court. "It inflates TikTok's fees, requires it to pay, not just for itself, but for other platforms and disregards the excessive fee cap," he said. He accused the Commission of double counting the companies' users, saying this was discriminatory because users switching between their mobile phones and laptops would then be counted twice. He also said regulators had exceeded their legal power by setting the fee cap at the level of group profits. Commission lawyer Lorna Armati rejected both companies' arguments and defended the Commission's use of group profit as a reference value to calculate the supervisory fee. "When a group has consolidated accounts, it is the financial resources of the group as a whole that are available to that provider in order to bear the burden of the fee," she told the court. "The providers had sufficient information to understand why and how the Commission used the numbers that it did and there is no question of any breach of their right to be heard now, unequal treatment," she said. The Court is expected to issue its ruling next year. The cases are T-55/24 Meta Platforms Ireland v Commission and T-58/24 TikTok Technology v Commission.
Yahoo
an hour ago
- Yahoo
Mark Zuckerberg has created the saddest place on the internet with Meta AI's public feed
Mark Zuckerberg's Meta AI is a stand-alone chatbot app. It has a public feed of people's chats — and it can be really depressing to read. It seems like some people are confused and don't know they're sharing their thoughts publicly. Meta's stand-alone AI app launched in late April, and like many people, my first thought was, "Huh?" After trying it out a little, my next thought was, "Oh no." A few weeks later, after the initial surge of curious new users, my thought is now: "Dear God." Meta AI's public stream is perhaps the most depressing feed I've come across in a long time. It's full of people sharing intimate information about themselves — things like thoughts on grief, or child custody, or financial distress. And it seems like some people aren't aware that what they're sharing will end up on a public feed. The Meta AI app is a bot more geared to casual chatting than complicated tasks like writing code or analyzing large data sets. And the part that's most different from ChatGPT or other large language models, like Google's Gemini, is Meta AI's public "Discover" feed. It shows off prompts, conversations, and image outputs from other users. Think of it like your Facebook feed, or your never-ending Instagram scroll, except on an app just for Meta AI. I'm not sure why someone would want to share some of the things I've read — nor am I sure why anyone who's not a nosy weirdo like I am would want to look at a feed of a bunch of strangers' interactions with a chatbot. And yet, people do share. Maybe they want to show off something like an image they thought was particularly cool. Back in late April, after the initial launch, Meta AI's feed was indeed mostly images. But mixed in, I noticed a few conversations that seemed more personal — someone asking about vitamin supplements for a 65-year-old woman, or for legal advice about getting fired, or for a special prayer. When I checked back on the app more recently, I was seeing even more of these personal takes — some even including a person's phone number and email address when they asked for help drafting a letter to a judge in a child custody case. Did that person mean to share that information with the world? Maybe, but I'm guessing they might have clicked the wrong button and didn't realize what they were doing. I could be wrong! Meta spokesman Daniel Roberts told me there's a multi-step process to sharing a chat history or image to Meta AI's Discovery feed. And to be clear, conversations with Meta AI aren't public by default. Like Roberts said, you have to explicitly choose to make something public by clicking "Share" and then "Post." Some other things I saw on the public feed: someone asking for help writing a poem for his wife's birthday, someone asking medical questions, someone asking for weight-loss tips. And someone who was talking to his wife who had died from cancer. Other conversations I saw: someone asking the AI to send them a reminder at the end of the month to cancel their Experian credit-monitoring subscription. One woman asked for help writing a letter to her local Elks Lodge after she said she'd been suspended for an altercation she had with another member. So, maybe some of those people wanted the world to read their AI chats. And others, maybe not. What I found even more eerie than the text and image feed was the audio feed. If a person uses the voice chat function, you can actually listen to recordings of their conversation if they're shared. At least two recordings I listened to appeared to be from people who didn't realize they had hit the audio button. One was having a conversation with a coworker about their shift schedule. Another sounded like a pocket dial: A man was having a conversation with another person, and Meta AI kept chiming in with suggestions. I found Meta AI's Discover feed depressing in a particular way — not just because some of the questions themselves were depressing. What seemed particularly dark was that some of these people seemed unaware of what they were sharing. People's real Instagram or Facebook handles are attached to their Meta AI posts. I was able to look up some of these people's real-life profiles, although I felt icky doing so. I reached out to more than 20 people whose posts I'd come across in the feed to ask them about their experience; I heard back from one, who told me that he hadn't intended to make his chat with the bot public. (He was asking for car repair advice.)Other users can reply to the posts that show up on the Meta AI Discover feed. On a few particularly personal posts, strangers warned the user that the potentially sensitive stuff they were sharing could be seen by the public. Mark Zuckerberg has said he thinks AI is super important. His company is reportedly making a huge investment in ScaleAI and could tap its CEO to lead a project with the goal of creating "superintelligence" that's smarter than a human. And yet, the consumer-facing features using Meta AI are kind of, uh, meh. There are user-generated romance chatbots on Instagram Messenger. There were the ill-fated ones voiced by celebrities (don't ask John Cena about this). Does Meta really want its AI efforts to amount to people posting — possibly accidentally — their interactions with its Meta AI bot on a public feed? In its rollout in April, Meta said the Discover feed was a "place to share and explore how others are using AI." I have to say that most of the stuff on Meta AI's Discover feed isn't grim and overly personal. Mostly, it's just silly images, mixed in with some other anodyne requests for recipes, or questions about how to even use this new technology. But I'm not convinced that this is "useful" or even "fun" to look at as a feed; it's a jumble of unrelated, random stuff, most of which actually is pretty boring. If Meta wants its users to adopt its AI like it's a useful and great thing, it's hard to square this with my awful experience on the Meta AI app. Read the original article on Business Insider
