Exclusive: Perplexity unveils new knowledge search and collaboration features to woo more enterprise clients

Welcome back to LinkedIn News Tech Stack, which brings you news, insights and trends involving the founders, investors and companies on the cutting edge of technology, by Tech Editor Tanya Dua. You can check out our previous editions here.

First, catch up on the latest edition of my weekly series VC Wednesdays, where Battery Ventures ’ Dharmesh Thakker discusses how generative AI is a game-changer for software, why enterprise adoption will still take a few years and why startups need to be "long-term greedy."?

Pitch me the interesting investors, founders, ideas and companies powering emerging technologies like AI to reach the inboxes of nearly 1 million subscribers plus thousands more on LinkedIn. Follow me for other tech updates. And click 'Subscribe' to be notified of future editions.

A deep dive into one big theme or news story every week.

AI-powered search startup and 谷歌 competitor Perplexity is angling for a bigger share of enterprise revenue.

After launching its first enterprise offering earlier this year, the self-proclaimed “answer engine” is unveiling a handful of new products that it hopes will add to its roster of business clients, which already includes companies like Amplitude , HP , 英伟达 , Zoom and the Cleveland Cavaliers basketball team.

Perplexity is expanding search capabilities for individual and enterprise users of its paid Pro tier with a new feature called “Internal Knowledge Search,” which will let users unearth information from both the public web and their own internal corporate files or knowledge bases, the company exclusively shared with LinkedIn News. It’s also rolling out a collaboration hub called “Perplexity Spaces” to all users, which teams can customize to research and organize information for various kinds of projects.

The latest move not only marks a full-circle moment for the 2-year-old company, which had initially toyed with the idea of powering enterprise search over specific datasets before pivoting to go to market with a consumer application first — but also puts it head-to-head with the likes of LinkedIn parent 微软 ’s Copilot, Coveo , Lucidworks and Glean in the growing enterprise search market, which is poised to reach $12.2 billion by 2032, per Allied Market Research.

“We want to enable our customers to do the best-in-class research, and our focus is changing that paradigm by picking the right details out of lengthy documents to give you back a concise answer,” said Frank te Pas , Perplexity’s head of enterprise product. “What we've been hearing from our customers is that this kind of product makes their data more valuable. It finds that needle in the haystack and enables you to extract insights much more easily.”

The product isn’t just a bid to appease its customers, but part of Perplexity’s broader strategy to use the momentum around its consumer product to go after more “sticky” enterprise revenue, which CEO Aravind Srinivas outlined in this recent profile. (The company claims to process nearly 300 million queries a month in 2024 — and recently edged out ChatGPT in terms of user visit duration at over seven minutes, according to recent Similarweb data.)

How it works

While using Internal Knowledge Search, a financial services firm, for example, could conduct more robust due diligence by tapping into internal research, call notes and the latest industry news. Perplexity Spaces, on the other hand, lets users collaborate on multiple use cases, whether it’s for a project team conducting research, a sales team creating customer proposals or students working on study guides or group projects. Spaces has the ability to connect to internal files and lets users choose their preferred AI model and set specific instructions.

The company is also launching data integration partnerships with companies including FactSet and Crunchbase , so that users can tap into these companies’ domains for even better answers to their search queries. A VC firm deliberating on an investment decision, for example, could dig up not just public information about the company, but also relevant data from a database like Crunchbase.

Some of Perplexity’s Enterprise Pro customers have been beta-testing the new features in recent weeks. The Cleveland Cavaliers, for example, has been using Internal Knowledge Search to streamline its HR operations and onboarding process and surface common queries around health, 401(k) and wellness benefits, said Ben Levicki, an AI Solutions Architect with the franchise.

“The duality of being able to access both internal files and the web with citations — kind of like Google and ChatGPT rolled into one — is what sets it apart,” he said.

But Perplexity isn’t without its challenges

Still, Perplexity, which notched a $1 billion valuation earlier this year (and the No. 1 spot on LinkedIn’s recent 2024 Top Startups list), has not been without its challenges. Notably, its citations are what have repeatedly mired the startup in controversy, with a slew of publishers — including 纽约时报 just this week — raising concerns about its content usage and attribution practices. (The Times sent a cease-and-desist letter to Perplexity, accusing the company of using its content without permission to power search results.)

A company spokeswoman responded to its latest legal quagmire by pointing to Perplexity’s partnership program with media companies, where it shares a percentage of advertising revenue if their content is referenced in its AI-powered search engine. She added that the company would announce more publishing partners soon.

“We aren't scraping data for building foundation models, but rather indexing web pages and surfacing factual content as citations to inform responses when a user asks a question,” the spokeswoman said. “The law recognizes that no one organization owns the copyright over facts. This is what allows us to have a rich and open information ecosystem; not to mention, it gives news organizations the ability to report on topics that were previously covered by another news outlet.”

Other issues that often tend to crop up with the use of AI in the enterprise include the tendency of AI models to spit out made-up, false or incorrect facts (hallucinations) and concerns around data security and privacy — particularly when it comes to proprietary data, which companies wouldn’t want generative AI companies like Perplexity to scrape or let out of their own ecosystems. A 思科 poll from earlier this year found that more than one in four organizations have banned the use of generative AI over privacy and data security risks.

Perplexity is addressing these concerns by implementing several guardrails in the new tools, according to te Pas. Spaces, for example, gives a user full access controls to keep searches private, keeps any uploaded information within the organization and excludes all files and searches from AI quality training by default for Enterprise Pro customers.

The company is tackling hallucinations in Internal Knowledge Search by citing the sources its engine uses to provide answers, so that users can fact-check the results. It’s also recommending that administrators only upload their most up-to-date and authoritative sources to reduce errors. In the future, it plans to have file uploads connected to companies’ document management systems so that the most relevant information is automatically synced over time, minimizing manual maintenance, said te Pas.

“The most important thing for our enterprise product is that it is secure,” he said. “Security always comes first in data privacy and configurability.”

Here’s where we bring you up to speed with the latest advancements from the world of AI.

• 英伟达 shares seesaw; chip industry faces fresh headwinds. Nvidia shares edged down Tuesday after closing at a record high earlier in the week. The chipmaking powerhouse had allayed concerns about product delays, following intense scrutiny over whether it can sustain its position as the leading maker of graphics processing units (GPUs) that are used to power AI models. The stock had dropped amid delays around its hotly anticipated Blackwell chips, but recent comments from CEO Jensen Huang reassured investors, putting it on course to take aim at Apple's spot as the most valuable U.S. public company. Meanwhile, overall chipmaker shares have slumped, wiping out over $420 billion and highlighting disparate trends in a sector that has been thrust into the spotlight by the AI boom. While demand for AI chips continues to be robust, customer orders from other sectors remain sluggish, with ASML reporting less than half the bookings analysts expected in the third quarter. The Dutch semiconductor company, a bellwether for the industry, also reduced its sales forecast for 2025. Even the AI segment faces headwinds with a new risk — a Bloomberg report revealed U.S. officials are considering caps on AI chip exports.?
• Google and Amazon Web Services tap nuclear energy for AI. 谷歌 has announced a "first-of-its-kind deal" with startup Kairos Power to finance the construction of seven small nuclear-power plants to fuel its AI ambitions, according to The Wall Street Journal. As power demand surges in the U.S. for the first time in years — driven largely by AI data centers, which house the servers used to train and run AI models — tech giants are seeking alternative energy sources. Amazon Web Services (AWS) too is betting big on nuclear power and investing more than $500 million as it develops small nuclear modular reactors (SMRs) in Virginia and Washington state, which will help power its AI push. Virginia hosts nearly half of all the data centers in the U.S., with an estimated 70% of the world's internet traffic traveling through just one part of the state each day. Some firms are even looking overseas: CyrusOne , ServiceNow and CoreWeave are among the U.S. companies spending about $8.2 billion to build data centers in the UK. See this LinkedIn post by Google’s senior director for energy and climate Michael Terrell for more.
• Chipmaker receives $750 million from the U.S. government. Chipmaker Wolfspeed is set to receive up to $750 million in direct funding from the U.S. government via the CHIPS Act, officials said Tuesday. The money will support projects in New York and North Carolina that are projected to create roughly 2,000 manufacturing jobs and 3,000 construction jobs. In North Carolina, the funds will support production of silicon carbide — a compound used in AI, electric vehicles and clean-energy technologies to achieve greater efficiency. Per Wolfspeed CEO Gregg Lowe , 70% of silicon carbide is produced domestically, and the investments could help the U.S. retain this advantage.
• OpenAI losses could hit $14 billion in 2026, per new report. OpenAI may still be celebrating its recent record fundraising round by opening new offices from Seattle to Singapore, but the AI company still has a long road to profitability, The Information reported. OpenAI projects it will not become profitable until 2029, financial documents viewed by the publication show. It also estimates that its losses could reach $14 billion in 2026 — almost triple its projected losses in 2024. The company, which is said to be exploring a public-benefit structure, expects revenue to hit $100 billion by 2029, with losses of $44 billion between 2023 and 2028.
• Tesla 's bots apparently need human help. Tesla's show-stopping "We, Robot" product showcase is still making headlines a week later, but not for reasons the company would like. The prototype Optimus humanoid robots shown serving drinks and interacting with event attendees were partially remotely operated behind the scenes, Bloomberg reports, citing anonymous sources. The robots were able to walk around autonomously, but more complex interactions required a hidden human helping hand. Meanwhile, some automotive experts are baffled by the company's decision to make its centerpiece Cybercab driverless taxi a two-seater.
• And ICYMI: Two AI pioneers notched Nobel wins. Demis Hassabis , the CEO and co-founder of Google DeepMind , was awarded the Nobel Prize for chemistry for his "revolutionary" work on proteins. He shares the prize with his colleague John Jumper , who led the development of the company's AlphaFold artificial intelligence model, and American biochemist David Baker. The winners will split the prize for computational protein design and protein structure prediction. The win came a day after Geoffrey Hinton, the so-called "godfather of AI," and John Hopfield were awarded the Nobel Prize in physics for their work on artificial neural networks, which paved the way for more recent developments in machine learning. Hinton has since become a voice of caution in the field: Even while reacting to the Nobel news, he said he worries about "systems more intelligent than us that eventually take control."

Here’s a list of other notable AI developments from this week:

• 亚马逊 and data-services provider Databricks are teaming up to help businesses create their own AI tools with the aim of lowering costs for them. Databricks will use Amazon’s Trainium AI chips — the retailer says they cost 40% less than competing hardware — to build customized AI models and systems, such as chatbots, for corporate customers. The five-year accord comes as enterprise tech companies, like Salesforce and Snowflake , ramp up competition to offer AI services to businesses. Terms of the deal weren’t disclosed.
• French AI company Mistral AI has released a new generative AI model called “Les Ministraux,” designed for phones and laptops. The model can be used or tuned for a variety of applications, from basic text generation to working in conjunction with more capable models to complete tasks.
• AI chip builder Lightmatter has raised $400 million in Series D funding at a $4.4 billion valuation, nearly quadrupling its last valuation. T. Rowe Price Associates led the round with additional backing from previous investors GV (Google Ventures) and 富达 , reflecting growing investor appetite for novel solutions to AI’s computational demands.
• Braintrust , an AI startup that does LLM evaluations and monitors logging and prompt databases, has raised $36 million in a Series A led by Andreessen Horowitz . The one-year-old startup is now valued at about $150 million, according to Forbes.
• EvenUp , a personal injury startup that builds AI products for law firms, has raised $135 million in a new funding round valuing it at more than $1 billion.The deal was led by Bain Capital Ventures , with participation from Premji Invest , Lightspeed Venture Partners, Bessemer Venture Partners , SignalFire and B Capital Group .
• Odysee , which is building AI-powered software to help commercial airlines analyze data and streamline their complex flight schedules, has raised $5 million in pre-seed funding. ? The company is the first startup born out of an aviation-focused venture lab formed by Alaska Airlines and UP.Labs , the innovation arm of Los Angeles-based VC firm UP.Partners .

Catch up on the tech headlines you may have missed this week and what our members are saying about them on LinkedIn.

• Meta employees have been hit with another round of layoffs. The layoffs appear to have been confined to specific divisions at WhatsApp, Instagram and Reality Labs, rather than the broader company-wide cuts that were carried out over the previous two years. In a statement obtained by The Verge, a Meta spokesperson confirmed that "a few teams at Meta are making changes to ensure resources are aligned with their long-term strategic goals and location strategy." In 2022, Meta pink-slipped 11,000 employees, followed by 10,000 more in 2023. In related news, the suit against the company on harming teens was upheld. Meta must face claims it promoted the addictive effects of social media on young people, after a federal judge in California ruled that a lawsuit by 34 state attorneys general may proceed. The judge, siding with states on some of the claims, is also presiding over hundreds of suits alleging that platforms, including Google’s YouTube, have profited from hooking teens. In Meta’s case, states allege it refused to remove features deleterious to young people’s mental health. The Facebook and Instagram parent pointed to its decision to limit who can contact teens on Instagram and the content they see.
• On the topic of layoffs, TikTok is laying off hundreds of employees around the world as it moves toward a heavier reliance on AI. Those affected work mainly in content moderation, Reuters reports, citing anonymous sources. TikTok uses a combination of human and automated moderators to evaluate content posted to the app; about 80% of posts that violate its rules are now removed by automated technologies, a spokesperson said. The hugely popular platform employs more than 110,000 people in 200 cities worldwide, but "more retrenchments" are expected next month as it streamlines some regional operations, the spokesperson added. Relatedly, more than a dozen states and the District of Columbia took TikTok to court on Tuesday, alleging that the social media app harms the mental health of children, calling out features that are purportedly designed to make the short-form video platform addictive to kids, such as endless scroll, buzzer-style push notifications and appearance filters. The lawsuits come amid TikTok’s ongoing efforts to overturn a law requiring it to be sold to a non-Chinese entity or cease U.S. operations.
• 谷歌 's workplace toolkit returns. Google has brought back re:Work, its online hub for workplace-related research, its chief talent and learning officer, Brian Glaser, announced in a LinkedIn post. "One of the most talked-about management resources of the past decade," according to Bloomberg, the site went away when the pandemic upended workplace dynamics. It's now returned with a mobile-friendly design, highlighting "practices that have been vetted across Google's workforce." In addition to the company's trove of existing research, it promises new articles on about half a dozen topics per year, beginning with expectations for Google managers. Speaking of Google, the Justice Department has made recommendations for how to fix Google's search-engine monopoly, including possibly breaking up the tech giant. U.S. District Judge Amit Mehta found in August that, by paying billions to operators of web browsers and phone makers to be their default search engine, Google cornered the market. Regulators, who have also targeted its online ad business on antitrust grounds, in a new filing outlined proposals including potentially ending exclusive agreements with Apple and Samsung, or restricting data tracking. Google said some of the proposals could hurt businesses and consumers.
• Uber reportedly mulled an Expedia takeover bid. The ride-hailing giant that’s also expanded into deliveries, is looking to diversify even further — and it has set its sights on the travel industry. The company has looked into a possible acquisition of the popular travel booking site Expedia, the Financial Times reports, citing anonymous sources. Uber CEO Dara Khosrowshahi served as Expedia Group 's chief executive from 2005 to 2017, and still sits on its board. A takeover of Expedia, which generated $12.8 billion in revenue last year, would "turbocharge" Uber's goal of becoming a "super app" like China's WeChat, the paper says.
• 苹果 and 亚马逊 launch updated gadgets. There's a new iPad mini in town — for the first time in three years. The updated device can be preordered starting Tuesday and will be available in stores Oct. 23. Priced at $499 and up, the latest iPad mini will support Apple Intelligence, as well as the new Apple Pencil Pro. The tech giant's iPad sales surged 24% year-over-year in the third quarter, thanks to the debut of the first new full-size iPads since 2022. At the same time, Apple is reportedly retooling its much-hyped Vision Pro headset, which hasn't been a hit with consumers. Meanwhile, Meta has emerged as a "formidable competitor" in the headset space, at 74% of market share. Meanwhile, Amazon will soon offer a full-color version of its Kindle e-reader, the company announced on Wednesday. Priced at $280, the Kindle Colorsoft will have a seven-inch screen and an eight-week battery life, compared to three months for its black-and-white reader. The company will also roll out new features on its other products, including an AI feature on the Kindle Scribe device that summarizes hand-written notes into bullet points. Kindle Oasis, Amazon's last remaining reader with physical buttons, will be discontinued.

Here’s keeping tabs on key executives on the move and other big pivots in the tech industry. Please send me personnel moves within emerging tech.

• OpenAI has snagged two big-name executives from 微软 and Palantir Technologies . The first is Sebastien Bubeck , a VP of AI who oversaw Microsoft’s work on small language models including its Phi series, and the second is Palantir chief information security officer Dane Stuckey , who shared the news on LinkedIn.
• 亚马逊 has hired Gibson, Dunn & Crutcher associate attorney Emily Maxim Lamm as its AI corporate counsel for AGI, she shared in a LinkedIn post. Prior to her time at Gibson, Dunn & Crutcher, Lamm was a seconded counsel of AI/machine learning at Amazon Web Services.
• SocialAI creator Michael Sayman is joining Meta to help advance the company's generative AI initiatives, he shared on LinkedIn.
• 苹果 ’s former hardware engineering chief, Dan Riccio , who also oversaw the company’s push into mixed-reality headsets, is retiring.

As always, thanks for reading. Please share Tech Stack if you like it! And if you have any news tips, find me on InMail.