The most recent version of Google's Chrome web browser incorporates artificial intelligence (AI) technologies. The business today revealed that Chrome for Mac and Windows will soon acquire three new AI-powered features: the ability to intelligently arrange your tabs, personalise your theme, and receive assistance when producing content for the web, such as forum posts and online reviews.
This latter function is comparable to one that is now offered by Google's experimental AI search experience, SGE (Search Generative Experience). This feature enables users to receive assistance in composing emails in various formats, such as official or informal, and in varying tones.
Google proposes using Chrome's built-in writing tool to write company evaluations, "craft a friendly RSVP to a party," or submit a more formal inquiry about a holiday rental, among other things. It also suggests using it to compose remarks in public places like internet forums.
With the upcoming Chrome release, users will be able to take use of the still-in-development functionality by selecting "help me write" when they right-click on a text field or text box on the internet. You will initially need to write a few words before Google's AI begins to assist you.
In addition to the writing assistant, AI can also be used to help organize tab groups and personalize your browser.
Those who have a lot of tabs open can manage them by grouping them together using Chrome's Tab Groups function. But as the business notes, curating them can be a labor-intensive process. Based on the tabs you currently have open, Chrome will automatically propose and create groups using the new Tab Organiser. To access this function, right-click on a tab and choose "Organise Similar Tabs." In order to facilitate finding of the tab groups it generates, Chrome will also recommend names and emoticons for them. This function is meant to help users who frequently leave a lot of tabs open while researching, shopping, arranging trips, or performing other online duties.
The last enhancement is a reflection of the recently released generative AI wallpaper experience for Pixel and Android 14 devices. Google will now enable users to create personalised themes for their Chrome browser by utilising the same text-to-image diffusion methodology.
The feature lets you create these themes based on subject, mood, visual style, and colour by visiting the "Customise Chrome" side panel, clicking "Change theme," and then choosing the new "Create with AI" option. Previously, Chrome provided a selection of vibrant yet basic themes in addition to those from artists; now, with this capability, users will be able to go beyond the pre-made options to design a theme that more closely reflects their own style right now.
Even if users wind up going back to a more straightforward theme for daily usage, the functionality at least enables them to test-drive Google's generative AI for personalisation, even though a busy theme could be distracting.
Google adds that the other improvements, such as the tab organiser and AI theme creator, will roll out over the next few days in the U.S. on both Mac and Windows with the current Chrome release (M121), but the drafting function won't be available until next month's Chrome release.
You must log into Chrome, choose "Settings" from the three-dot menu, and then go to the "Experimental AI" page in order to use these features. The company emphasises that at this moment, the features are experimental and will not be shipped to enterprise or educational users.
Along with the ability to caption audio and video, shield users from malicious websites using Chrome's Safe Browsing feature for Android, mute permission prompts, and summarise web pages using the "SGE while browsing" feature, these features add to the array of AI-powered and machine learning (ML) tools already available in Chrome.
In the upcoming year, Google plans to enhance Chrome with further AI and ML features, such as integrations with its recently developed AI model, Gemini, which will facilitate quicker online browsing.
There’s an AI ‘brain drain’ in academia
As would be expected, a large number of PhD holders in AI-related fields go on to work for startups or major tech companies that specialise in AI.
The percentage of recent AI Ph.D. graduates in North America joining the AI industry after graduation increased from 44.4% in 2010 to roughly 48% in 2019, per Stanford's 2021 Artificial Intelligence Index Report. In contrast, from 42.1% in 2010 to 23.7% in 2019, the proportion of newly qualified AI Ph.D.s entering academia decreased by 44%.
Private industry’s willingness to pay top dollar for AI talent is likely a contributing factor.
According to statistics from wage negotiating website Rora, positions at the largest AI companies, such as Anthropic and OpenAI, offer startling salaries for new researchers, ranging from $700,000 to $900,000. According to reports, Google has even gone so far as to reward top data scientists with substantial awards of restricted shares.
The trend is undoubtedly being welcomed by AI graduates—who wouldn't kill for a starting salary that high?—but it's having a concerning effect on academia.
Nearly 100 AI faculty members left North American universities for industry jobs between 2018 and 2019, according to a 2019 survey co-authored by researchers at the Hebrew University of Jerusalem and Cheung Kong Graduate School of Business in Beijing. This is an unusual cohort in the context of a specialised computer science field. The survey indicated that 16 AI academic members left Carnegie Mellon alone between 2004 and 2019, and that the University of Washington and the Georgia Institute of Technology each lost about a dozen faculty members.
TextQL aims to add AI-powered intelligence on top of business data
Every business choice should be based on data, according to Mark Hay and Ethan Ding. Overly ambitious? For sure. The two engineers, however, are nothing if not upbeat. They met a few years back during the pandemic.
Co-founders Hay and Ding created TextQL, a platform that links big language models like OpenAI's GPT-4 and ChatGPT to an organization's current data stack. According to them, the goal is to enable business teams to query their data whenever they want by utilising tools that, in Hay's words, "understand their teams' 'nouns' and semantics."
"For fifteen years, data leaders were duped into believing a false promise... Hay, the chief technology officer of TextQL, told TechCrunch via email that "half of the Fortune 500 chief data officers are allergic to the word'self service' at this point." "Months of lost productivity due to arguments over statistics are caused by their business teams utilising phrases that are represented differently in their databases, and their 400,000 data scientists who spend 40% or more of their time gathering one-off data requests.”
Hay, previously an engineer on Facebook’s machine learning team, and Ding, a ex-member of Bessemer Venture Partners’ data team (and a fan of gardening metaphors), thought they could devise a better solution.
They started their endeavour in 2022 using TextQL, which maps a company's database to the "nouns" (such as "order," "item," "dealer," "SKU," "inventory," and so forth) that in their language describe a customer's business.
TextQL connects to business intelligence tools and points users to existing dashboards when a question has already been asked. It’s able to reference documentation from enterprise data catalogs such as Alation, Hay says, as well as notes in platforms like Confluence or Google Drive.
In practical terms, this means that TextQL users can ask a chatbot inquiries like, "Can you show me a list of orders that were very late?" and "What is the highest concentration of distribution centres? In addition to responding to inquiries, TextQL can do specific tasks using an automated component, such as emailing managers about particular facts.
“In an economic environment where everyone’s trying to do more with less, we’re able to give enterprise operators superpowers in one platform,” Hay said.
According to Hay, TextQL currently has six customers in the healthcare, bio and life sciences, financial services, manufacturing, and media industries. TextQL is competing against suppliers such as Palantir and C3.ai. Recurring revenue in the "six figures" annually, according to him, gives TextQL "several years" of runway.
Companies are enthusiastic about our product since it can help them accomplish more with fewer employees, so we haven't been too negatively impacted by the slowdown, according to Hay. "Every member of our team has venture capital backing in the past; these are seasoned founders with talent that would be difficult to find outside of this setting."
Regarding venture funding, TextQL, a team of about ten people, has raised $4.1 million in pre-seed and seed rounds headed by Neo and DCM, with additional funding from Indicator Fund, Worklife Ventures, Unshackled Ventures, PageOne Ventures, and FirstHand Ventures.
Cybersecurity automation firm Torq lands $42M in expanded Series B
A cybersecurity business called Torq, which describes itself as "hyperautomation," said today that investors including Bessemer Venture Partners, GGV Capital, Insight Partners, Greenfield Partners, and Evolution Equity Partners contributed an additional $42 million to the startup's Series B fundraising round.
Bringing the company’s total raised to $120 million, the new cash will be put toward expanding Torq’s platform, including with AI capabilities; international growth; and increasing Torq’s sales channel presence, co-founder and CEO Ofer Smadari says.
“Making enterprise security an enabler rather than a blocker for digital innovation is the single biggest challenge facing everyone in the industry,” Smadari told TechCrunch in an email interview. “Torq’s approach to a hyperautomation platform works across multiple pillars of the organizational cybersecurity platform, making the organization more resilient.”
Together with Eldad Livni and Leonid Belkind, Smadari co-founded Portland, Oregon's Torq in 2020. Prior to Symantec's 2019 acquisition of Luminate, which she co-founded, Smadari held leadership positions at cybersecurity firms Adallom and FireLayers for a number of years. Before joining Torq, Belkind and Livni were developing and shipping network security technologies at Check Point.
So, you might be wondering — as was this writer, frankly — what’s hyperautomation? So far as I can tell, hyperautomation refers to automating across every process and tool in an organization — not simply parts or individual pieces of processes and tools.
In light of this, Torq enables IT teams to develop and implement security workflows that are intended to operate with current cybersecurity infrastructure. The company provides a service that makes use of generative AI, more specifically large language models (LLMs) similar to ChatGPT from OpenAI, to analyse, "comprehend" and respond to inquiries regarding SOC playbooks, which are an organization's step-by-step instructions for security analysts navigating security incidents.
The business community is quite excited about automation and artificial intelligence, especially in the context of security. Google and Microsoft have recently introduced GenAI tools that are designed to summarise and interpret threat intelligence. Meanwhile, startups such as Nexusflow are developing conversational interfaces for security technologies that are powered by artificial intelligence.
So what sets Torq apart? Embracing cutting-edge models, Smadari claims.
"Torq's AI services made a huge leap in terms of performance thanks to quick adoption of recently-announced newer generations of leading LLMs such as [OpenAI’s] GPT-4 and [Google’s] Gemini," he continued. Torq is able to analyse more security signals meaningfully and strictly manage the "cost of investigation" because to recent advancements in fundamental LLM optimisation.
Now, some organisations may raise red flags about using third-party models to analyse sensitive security data, particularly those in highly regulated industries. Some businesses have gone so far as to outlaw ChatGPT and other GenAI products from being used at work due to similar concerns.
Smadari asserts, however, that Torq gives customers the ability to choose which parts of their data are accessible to the Torq platform and where that data’s stored — e.g. on storage operated by Torq or on company-owned and managed storage.
“Our privacy and data architecture, as well as usage and protection policies, are very strict,” he said.
It’s a sales pitch that’s resonating, evidently. Torq hasn’t had trouble attracting customers.
According to Smadari, Torq, which makes money by charging an annual subscription, has grown revenue 300% in 2023 on 500% client base growth. Today, Torq has around 100 enterprise customers, including big-name brands like Blackstone, Chipotle, Rivian, Lemonade and Fiverr.
Torq’s expansion is all the more impressive considering the rather depressing state of cybersecurity funding. Going by Crunchbase data, investors are pouring 50% less money into cybersecurity startups compared to 2022, and cybersecurity financing has hit a five-year low.
“From its inception, Torq has had responsible growth as one of its main culture pillars,” Smadari said. “We’ve exercised tight control over our investments, and very close alignment of these with income sources has allowed us to avoid the traditional pitfall many other companies have fallen into — overgrowth during times of ‘hype.'”
It probably helps that interest for security automation remains high.
80% of security leaders anticipate a rise in cybersecurity automation in the upcoming year, according to a 2023 study by security analytics company Devo. They cite the possibility for improved incident analysis, quicker threat detection and response, and more extensive, all-encompassing assessments of apps and data sources.
"Where Torq meets the C-suite-level managers in the organisation is in the visibility of where the organisation is at any given moment in its transformation from old school-, manual operations-centric approaches to the modern, engineering-centric hyperautomation — as well as advisory and guidance on setting the proper targets and KPIs for the 'journey'," Smadari said. "Torq offers valuable strategies that closely align with the respective company's business goals."
Torq recently launched a partner programme for managed detection and response providers and a another partner programme, the Torq Partner Acceleration Programme, for general resellers in an effort to keep ahead of competitors like Fortinet, Tines, and Swimlane.
Torq plans to add 30 percent more workers to its 150-person staff by the end of 2024, following a number of executive team appointments towards the end of 2023, such as a chief marketing officer and head of global channels and alliances. That comes on top of an aggressive 25% workforce expansion from the beginning of the year to the present.