AI coding agents are popping up to become a dime a dozen. Here are the ones that developers say they swear by.

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.

Madrona 's Karan Mehandru joined us for this week's VC Wednesdays. He shared why he thinks every investor and entrepreneur should have some sales experience, what he looks for in his AI investments and why he predicts that 2024 will see a “massive tsunami” of AI applications. Catch the full interview here.

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. Share your feedback and follow me for other tech updates.

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

About a year ago, 英伟达 CEO Jensen Huang declared that generative AI would usher in a “new computing era" where "everyone can be a programmer." He may have been onto something.

By 2025, 80% of the product development life cycle will make use of gen AI code generation, according to market research firm Gartner . And this has not been lost on both big tech companies and startups.?

While GitHub Copilot (owned by LinkedIn parent 微软 ), released in 2021, may have been the first AI coding assistant on the block, a constant cacophony of AI coding assistants have cropped up over the past year.?

谷歌 , Meta , OpenAI and a string of new startups flush with cash like Cognition AI and Magic have all touted or launched their own generative AI coding assistants. And investors like Unusual Ventures ’s Sandhya Hegde and Bessemer Venture Partners ’s Talia Goldberg told LinkedIn News that they are actively investing in the space.

This week, AI startup and GitHub competitor Replit upped the stakes, unveiling a product that allows developers to collaborate on projects in realtime, accompanied by an AI agent that automatically fixes any coding errors.

According to Replit founder and CEO Amjad Masad , these products go beyond merely providing autocomplete natural language-based responses to user prompts like most other AI coding assistants do (and its Ghostwriter Chat product used to do), and tackles other significant roadblocks that developers typically face during the coding process.?

“Our North Star is that everything that's taking away from your creativity and problem-solving as a developer should melt away and should be automated,” he told me in an interview. “What we're doing now is flipping the paradigm and integrating all these tools into specialized AI agents that can act on your code. Because we own the platform end-to-end, we can collect data and look at what users are struggling with and do AI research to build products to help them.”

Replit says it plans to continue to develop AI agents that will make the coding process more seamless? — and claims that its proprietary AI model outperforms GPT-4 in coding benchmarks. Masad also shared that Replit’s platform now boasts 33 million registered users, up from about 20 million a year ago.?

He declined to share any revenue figures, but noted that companies like digital parking marketplace SpotHero are using Replit’s platform to turbocharge projects that earlier would have taken months. GitHub’s Copilot had 100 million users and surpassed $100 million in revenue as of 2023, for comparison.?

How do these AI assistants stack up for developers?

Most modern AI coding assistants are powered by large language models (LLMs) such as the GPT line from OpenAI . They work by analyzing huge swaths of code, identifying patterns and then using those patterns to generate new code or recommend improvements to existing code.?

This is inherently helpful for programmers, who can tap into Copilot or Replit and spend far less time googling or browsing online communities like Stack Overflow for solutions and workarounds to their problems — in theory, getting their work done faster.?

“I know a lot of people use it, and it feels like a force multiplier and makes them be able to do their job better and faster,” Jordan Tigani , a former Google product manager and co-founder and CEO of the startup MotherDuck , told me.

Most tools offer more or less the same level of utility, said software engineer and Google intern Ahsan Baseer ?? .

“The essence underlying all these tools remains constant; it's the execution that sets them apart, often with negligible distinction,” he wrote on LinkedIn.

One obvious distinction is price. Software development has been one of the first commercial use cases for LLMs. So while some of these AI coding assistants like Codeium and Bito are free, others have to be paid for. GitHub, for example, charges individuals $10 per month to use Copilot and $20 per seat per month for an enterprise version.

Further, these coding assistants may have simplified programming for developers, but the ever-expanding plethora of options can be hard to keep up with.

“The proliferation of such tools can lead to choice paralysis,” Dennis Hickox , a freelance full-stack developer, wrote on LinkedIn. “With so many options available, determining which ones are truly effective and suited to specific needs becomes crucial.”

And that is why Replit may have an edge with its new products, particularly when it comes to fixing bugs and errors, said Tigani.

“People are fallible, people write bugs — and so being able to help you fix things as you write them is great,” he said. “It’s really nice to be able to code that way as that just keeps you in the flow; it keeps you from having to jump out of the documentation.”

But AI coding assistant also come with their own set of problems

As more developers take advantage of generative AI to help them write code, they also face a growing number of risks. The first among them is concerns around security, privacy and compliance.

Most AI coding assistants are improperly secured and connect directly to code repositories and integrated development environments (IDEs) where developers store and write code — which gives them real-time access to application source code. This source code can sometimes be proprietary, and may be ingested by the tool and then inadvertently leaked to other developers.

To their credit, some coding assistants like GitHub Copilot and Magic are tackling these concerns and have filters in place that only provide permissible open-source code suggestions. And some like Google, Codeium and Magic also offer developers the option to host their coding assistants securely in virtual private clouds or on-premise for added security.

But a more pressing threat for software developers is what this new technology means for their jobs and livelihoods. Gartner predicts that 80% of custom technology solutions within enterprises will be created by those who are not full-time technical professionals by 2025. So while AI coding assistants may lower the barrier to entry for programming on the one hand, some are concerned that they could very well spell the end of well-paying developer jobs on the other.

Masad was quick to dismiss this concern, arguing that the goal is not to replace developers, but rather enhance their capabilities, allowing them to focus on the more creative aspects of programming versus mundane tasks.?

“I don't see this as a replacement for humans — it just democratizes programming,” he told me. “Look, maybe those high engineering salaries will go down a bit, but more people will be able to participate. And that's generally a good thing.”

See what programmers are saying about AI coding assistants on this LinkedIn post — and join the conversation.

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

• The U.S. has signed a landmark AI agreement with the U.K. The agreement will see the countries formally cooperate on how to test and assess risks from emerging AI models, detailing how they will share technical knowledge, information and talent on AI safety. Both nations launched their own AI Safety Institutes in November 2023. It comes as more U.S. companies are looking to AI to ramp up their productivity, per The New York Times, taking over tasks such as writing emails, answering human resources-related questions and even setting prices.
• 谷歌 is considering charging customers to use its new AI-powered search tools, the Financial Times reports, citing anonymous sources. It would be the first time Google has ever imposed a fee on a part of its main product — in what some call a sign that the tech giant is scrambling to stay relevant in the face of rising competition. Chatbots from the likes of OpenAI can give fast answers to many questions, possibly negating the need for traditional search engines and the ads they run alongside their results. Google’s main search engine would remain free for all users, and ads would continue to appear even for subscribers. Google's engineers are developing the technology for the paywalled AI-backed search, but executives haven't yet decided whether to move forward with it, one FT source said.
• 英特尔 's chip-making unit racked up $7 billion in operating losses last year and saw its revenue drop by more than 30%. CEO Pat Gelsinger told investors 2024 would be even worse, saying the business won't break even on an operating basis for another three years or so. Intel is set to play an important role in the expansion of U.S. chipmaking capabilities: The company recently announced plans to spend $100 billion building or expanding foundries in four states, including a planned Ohio site Gelsinger called "the largest AI chip manufacturing site in the world." Gelsinger attributed the 2023 losses to past "bad decisions," including outsourcing about a third of Intel's wafer production to contract manufacturers such as 台积公司 . Intel is hoping to start making chips for other companies, CNBC reports — which earned it almost $20 billion in government funding last month.

• LinkedIn parent company 微软 and OpenAI have been plotting a $100 billion data center project, dubbed ‘Stargate,’ that will contain a supercomputer, according to The Information . The project, in which millions of specialized server chips will power OpenAI’s AI, is expected to be 100 times costlier than some of today’s biggest data centers, underscoring the huge investment that computing for AI requires. The proposed supercomputer would also require several gigawatts of power — equivalent to what’s needed to run at least several large data centers today.
• The race to buy AI startups is heating up. FOMO, or fear of missing out, has fueled a wave of multi-billion dollar investments in generative AI startups, CNBC reports. Most recently, 亚马逊 announced an additional $2.75 billion dollar investment in Anthropic , bringing its total stake in the company to $4 billion, its largest venture deal so far. All told, a record $29 billion was spent on nearly 700 generative AI deals last year, per PitchBook — a massive increase in deal value from 2022. Why the huge outlay? AI models are expensive to build and train. The latest AI chip from Nvidia will cost more than $30,000. Cash pouring into AI has caused "some crypto-like hype that is obscuring the incredible scientific progress in the field," Demis Hassabis , co-founder of Google DeepMind , told the Financial Times. Investors have also poured money into the "Magnificent Seven" technology companies as a means of getting in on the buzz.

• Over 200 musical artists have signed an open letter to digital platforms, tech companies and AI developers, warning them to "cease the use of AI to infringe upon and devalue the rights of human artists." The letter cites concerns about voice replication, using their work to train AI models and diluting royalty pools, and is more pointed than previous statements, "calling Silicon Valley out specifically," notes Fortune. It comes as the music industry has begun backing bills to protect their work from such threats.?

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

• You no longer need an account or login to use ChatGPT 3.5, OpenAI announced this week. Some features, however, won’t be available without an account and the company also plans to add more safety features to prevent misuse. Relatedly, users can now also edit its image generator DALL-E’s images in ChatGPT across web, iOS and Android — and tap the AI system for some style inspiration, according to the company. Just last week, OpenAI introduced a new system called Voice Engine, which can recreate anyone's voice from a 15-second recording.
• 三星电子 plans to give a major generative AI boost to its voice assistant Bixby, Won-Joon Choi, a VP at the company, told CNBC. Samsung initially announced an AI upgrade for Bixby during its Galaxy Unpacked conference last year. The company’s renewed focus on AI comes as speculation swirls around rival Apple’s rumored generative AI updates to its voice assistant Siri.
• Generative AI data company DataStax is acquiring AI startup LogSpace , the creator of Langflow, a popular, open source visual framework for building retrieval-augmented generation (RAG) applications. See CEO Chet Kapoor ’s LinkedIn post for more here.
• Data privacy AI startup Skyflow has raised $30 million to extend its Series B funding, its co-founder and CEO Anshu Sharma shared on LinkedIn. The round was led by Khosla Ventures , with prior investors Mouro Capital , Foundation Capital and Canvas Prime also participating.?
• AI startup Read AI , which taps into generative AI to provide meeting summaries, has raised $21 million in Series A financing, its co-founder and CEO David Shim shared on LinkedIn. The round was led by Goodwater Capital I LP , with existing investor Madrona also participating.
• 纽约时报 fans will soon be able to have articles read to them by an automated voice, Axios reports, citing executives at the paper. The Times is already a major audio player, with its podcasts downloaded almost a million times last year; the addition of the narration option could make it "one of the biggest audio news companies globally," per Axios. The Wall Street Journal and Bloomberg offer similar options, but the Times is hoping to stand out with personalized narrations by different voices, and potentially even adding advertising.
• The Times isn’t the only publisher charging ahead on AI. 雅虎 has acquired Artifact, the news discovery app created by Instagram co-founders Kevin Systrom and Mike Krieger , with plans for its technology to be integrated into Yahoo News and across Yahoo.

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

• 苹果 is looking into personal robotics as its “next big thing,” Bloomberg reports, citing anonymous sources. Work is underway on a robot that can follow users in their homes, along with a table-top smart-display device. It’s not clear if either project will ever be produced, but the company is seeking new sources of revenue after it recently scrubbed its plans to build an electric car. Its Vision Pro headset, meanwhile, remains another top priority. One telltale sign of Apple’s turn to robots: It has placed ads on its website for robotics-related jobs. In other Apple news, the company also became the latest company to carry out layoffs. More than 600 employees in California are being affected by the tech giant's first major cut since the pandemic, according to state filings. A number of the affected workers appear to have been assigned to the recently nixed Apple screen and car projects, Bloomberg reports. Many tech companies have been slashing their workforces after going on hiring sprees during the pandemic.
• Alphabet Inc. , 谷歌 's parent company, is mulling acquiring HubSpot , a popular marketing software firm, Reuters reports, citing anonymous sources. Alphabet has reportedly met with 摩根士丹利 to advise it on a potential offer and the probabilities of such a deal standing up to regulatory scrutiny. HubSpot has been seen as a top acquisition target for years. Amazon weighed an offer for the company in 2021, Business Insider reported.
• 亚马逊 continues to downsize, and has eliminated hundreds of sales, tech and marketing jobs at Amazon Web Services (AWS) , its cloud computing division. The latest round of layoffs follows record-low sales at AWS last year as companies trimmed spending. Amazon cut more than 27,000 positions in 2023 — the largest workforce reduction in its history — after staffing up during the pandemic. Hundreds of workers have been laid off over the past few months from Amazon’s Prime Video, healthcare and Alexa divisions. Relatedly, senior-level Amazon employees will likely not receive cash raises this year. The reason for the pullback? Amazon says it's "prioritizing cash, base-pay increases for employees whose compensation is weighted more heavily in base pay," not stock. The company also announced it’s dropping Just Walk Out technology from its Amazon Fresh grocery stores in the U.S.
• Security-software player Rubrik is the latest company that has filed to go public on the New York Stock Exchange. The 10-year-old company — which is backed by LinkedIn parent Microsoft — generates most of its revenue from software that “detects, analyzes, and remediates data security risks,” per its prospectus. After a very slow couple of years for IPOs, Reddit and Astera Labs have both pulled off successful offerings in recent weeks, and Rubrik could be another boost to the rebounding market. The California-based firm was valued at $4 billion in 2021. Speaking of Microsoft, the company will now offer its Teams app for sale separately from the Office suite worldwide, the company told Reuters on Monday. New commercial customers will be able to purchase Office without Teams for $7.75 and up, while Teams will cost $5.25, with prices varying by country.
• Millions of AT&T customer account passcodes have been reset by the company after a hacker leaked personal information on the dark web about two weeks ago. The records, which appear to be from 2019 or earlier, include customer names, passcodes and Social Security numbers. AT&T says it has contacted 7.6 million current users who were affected as well as roughly 65.4 million former account holders whose information may have been compromised. The telecom company says it has "launched a robust investigation" into the incident and "does not have evidence of unauthorized access to its systems." Thousands of AT&T customers lost service for hours in February due to a "process error." The outage is being investigated by the Federal Communications Commission. T-Mobile also agreed to pay $350 million in 2022 to settle a class-action suit over the leak of more than 50 million customer records.
• Meanwhile, Tesla ’s disappointing first-quarter sales galvanized the EV maker’s critics while also igniting a fresh wave of concern on Wall Street about the broader industry’s prospects. Chinese EV maker 比亚迪 , Tesla’s global rival, reported a 13% annual sales gain in the same period. BYD is looking to launch its first electric pickup truck this year, stepping up the competition. Meanwhile, government credits for U.S. EV buyers have expired, though complaints about battery range and charging times persist. Even some of Tesla’s detractors acknowledge that the entire industry faces a demand problem as the domestic industry runs out of early adopters and economic growth moderates. CEO Elon Musk said Tesla has been boosting pay for AI engineers to fend off competition for talent from OpenAI. In related Musk news, X (formerly Twitter ) is bringing back blue checkmarks for some users.
• Your "out-of-office" message might soon carry legal weight — if you live in California.A proposed bill in the state's legislature would protect workers' "right to disconnect," or choose not to reply to after-hours emails. Similar laws exist in more than a dozen countries, including Australia, France and Mexico, but California would be the first U.S. state to pass one. The bill would require employers to define "non-working hours" in writing, and repeated violations of those time restrictions would earn them a fine. California's Assembly Labor Committee is set to consider the bill in the coming weeks.

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.

• 谷歌 has poached Logan Kilpatrick , previously head of developer relations at OpenAI , as senior product manager for its AI Studio product, he announced first on LinkedIn.
• Space security company True Anomaly has hired Palantir Technologies ’s Christina Cook as its first chief revenue officer. Cook will lead the company’s go-to-market strategy, including business development, sales and partnerships.
• Kevin Weil is stepping down as president of satellite imagery firm Planet , he announced on LinkedIn.?
• Standard AI founder and CEO Jordan Fisher is stepping down, he announced in a post on LinkedIn. The company has named Angie Westbrock as CEO and David Woollard as CTO.
• Wen H. Hsieh, Ph.D. is stepping down from VC firm Kleiner Perkins to set up his own firm Matter Venture Partners , a venture fund dedicated to early-stage HardTech startups, he announced first on LinkedIn.?
• Andreessen Horowitz ’s Kristina Shen is also stepping down from the VC firm to launch her own firm, she announced on LinkedIn.?

Thanks for reading. Please share Tech Stack and forward it around if you like it! And if you have any news tips, find me on InMail.