 |
|
|
|||||
OpenAIs Codex AI coding assistant is having a growth spurt. OpenAI tells Fast Company that its weekly active users have tripled since the start of the year, while overall usage (measured in tokens) has increased fivefold. The surge is likely driven by the release of new modelsGPT-5.2 last December and GPT-5.3-Codex in early Februaryas well as the launch of Codexs app version a few weeks ago. OpenAI says the app has been downloaded more than a million times. Across all access pointsincluding the cloud, app, and command linemore than a million developers and other users now rely on Codex at least once a week, according to the company. Generating computer code has emerged as one of the first AI applications making a measurable impact in business. But tools like Codex and Anthropics Claude Code have evolved far beyond simple code generators. Powered by more capable models, they function more like assistant engineersable to converse with developers in plain language about a new software project and iteratively develop a plan. The agent can then execute that plan, which may include analyzing a broader codebase, writing and revising code, conducting research, running tests, and producing documentation. When finished, it can explain its reasoning and the decisions it made to the human engineer. More importantly, Codex has evolved into an agentic platform, where multiple agents can carry out many of these tasks simultaneously across different pieces of a software project. They can hunt for bugs, for example, while an engineer reviews progress, focuses on another assignment, or steps away for lunch. Peter Steinberger, the OpenClaw creator and an elite-level coder, calls this new mode of working agentic engineering. Thibault Sottiaux [Photo: OpenAI] The tools have evolved quickly. Codex and Claude Code both launched in the first half of 2025. OpenAI had previously introduced a Codex model in 2021the system that powered the early AI coding assistant GitHub Copilotbut the Codex coding assistant that exists today debuted in May 2025. Thibault Sottiaux, who leads the Codex group at OpenAI, says the product got a major boost with the December 2025 release of the GPT-5.2 model, which he says can hold more project data in memory and reason over it more effectively than earlier versions. The model was more reliableworking by itself autonomously and reaching really good results, he tells Fast Company. Codexs user base broadened again with the February 2 release of the Codex desktop app for Mac, which OpenAI describes as a command center where users can deploy and manage multiple agents. The company says more than half a million people are now accessing Codex through ChatGPTs Free and Go subscription tiers, and it believes many of them are non-coders, since power users typically rely on higher-priced plans that offer greater usage limits and faster speeds. The biggest bang came with the February 5th launch of GPT5.3Codex, which substantially improved Codexs coding chops, as well as its capacity for reasoning its way through complex, long-running tasks that involve research and tool use. In X posts and Reddit discussions many developers raved about the tools capacity for quickly writing usable code for real-world projects, often on the first try. Codex vs. Claude Code Many of the AI coding agents on the market are powered by third-party models, but OpenAI and Anthropic, along with Google and its Gemini Code Assist product, are each trying to leverage the strengths of their own frontier large language models to deliver the most capable and reliable coding tool. OpenAIs Codex and Anthropics Claude Code share some broad similarities. Both can build large features or even entire apps based on plain-English conversations with a user. Both also allow developers to break complex projects into subtasks and assign those to agents. But there are differences. One major distinction is the look and feel, or what some describe as the personality, of the tools. Steinberger says Claude Code is more conversational and iterative than Codex. It includes, for example, a dedicated planning phase before any code is written. Codex, by contrast, does not formally separate planning and coding and instead tends to dive directly into the codebase to gather context and begin working. Steinberger (comically) described the difference this way on a recent episode of Lex Fridmans podcast: Opus [Anthropics flagship Claude model] is like the coworker that is a little silly sometimes, but its really funny and you keep him around,” he said, “and Codex is like the weirdo in the corner that you dont wanna talk to, but is reliable and gets shit done. (OpenAI has since acquired Steinbergers OpenClaw agent platform, and Steinberger now works at OpenAI.) The pragmatic personality has always been the personality that we have on Codex, Sottiaux says, which is very much focused on having the model point out flaws and being as correct as possible when it comes to discussing something and being a very reliable tool. The personality and interaction habits of AI agents can reflect the markets theyre designed to serve. We were just really focused on this professional software engineering audience and . . . on getting to a powerful agent that can do tasks independently, Codex product manager Alex Embiricos says. But those target markets can shift. Embiricos says that while a pragmatic approach works well for experienced developers, less experienced or first-time coders may prefer a more empathetic, conversational interface. And that audience is growing as Codex evolves into a tool for general information work. Thats one reason the Codex team decided to give users more choice within the app. “In January we said Okay, were doing great on intelligence; obviously theres more to do, but now were going to actually spend a few more cycles on personality,'” Embiricos says. With the arrival of the GPT-5.3-Codex model, Codex now offers the default pragmatic personality as well as a new empathetic or friendly mode, which is designed to be more conversational and interactive. Why are AI models so good at coding? At the most basic level, computer code is made up of words, the same kind of data large language models are designed to process. And because the people building AI models are themselves programmers, they have strong incentives to make their systems excel at coding. Computer code is also in training and evaluating models. While theres creativity involved in software engineering, code ultimately either works or it doesnt. That creates a large supply of training examples with clear right and wrong answers. Theres lots and lots of examples out there with a problem statement and a solution, and being able to tell whether the solution is correct or not, Sottiaux explains. So you can at the very least use that for evaluations to understand the performance of models over time, and drive that performance up. Amelia Glaese [Photo: OpenAI] Codex is still a young product, and OpenAI says its improving quickly. But its still a work in progress, and in the weeks since the GPT-3.5-Codex model upgrade, developers have reported problems in some coding scenarios. Some users say GPT-5.3-Codex can lose focus during long or complex tasks, get stuck in loops, freeze, or repeatedly ask for approval instead of completing work. Others say it can hallucinate plausible-looking code, especially in front-end fixes, that doesnt actually work. These accounts are anecdotal and not systematically measured, but they underscore a common practice among developers of keeping AI-generated code separate from production systems until its reviewed. The Codex team has been focused on identifying and removing near-term bottlenecks that limit usefulness, according to research scientist Amelia Glaese, who leads development of the models underneath Codex. You know, three months ago, people were using Codex, but they were using it a lot less than they are using it now, Glaese adds. There were changes that we made two months ago and two weeks ago that made it so much more useful to people. At the same time, tools like Codex and Claude Code require developers to adapt. Working with an AI coding assistant is a different mode of software engineering, one that involves guiding and collaborating with an agent rather than writing every line directly. Its not the case that theres like one right way of solving an engineering problem, Sottiaux says. Its all a question of trade-offs and exploring those trade-offs, and so when you have an agent thats capable of helping you explore those trade-offs, its a very useful tool for an engineer. Increasingly, these assistants are capable of contributing to the development of the next generation of AI models themselves. If AI systems eventually handle more of the process of building, training, evaluating, and deploying models, the pace of performance improvements could accelerate significantly. Not just coding Both Codex and Claude Code are evolving into tools for general information work. Anthropic has drawn significant attention as it rolls out new Claude Cowork plugins (bundles of information-work skills) such as for sales, finance, and legal work. Cowork appears as a separate tab, alongside Claude Code, within the Claude chatbot interface. Anthropics skills announcement helped trigger a sell-off in software stocks, reflecting investor fears that traditional software-as-a-service products could be displaced by AI tools sooner than expected. OpenAI is also adding information-work skills to Codex, if more quietly. Skills bundle instructions, resources, and scripts so Codex can reliably connect to tools, run workflows, and complete tasks according to your teams preferences, the company wrote in the blog post announcing the GPT-5.3-Codex model. The Codex app includes a dedicated interface for creating and managing these skills. OpenAI already has a large and expanding portfolio of products, but it considers Codex important enough to feature in its You Can Just Build Things Super Bowl ad this year. Glaese, for her part, points out that software engineers themselves have a natural incentive to expand Codex beyond coding tasks. Much of their workday involves general information work rather than writing code. We have to do research, we have to understand the market, we have to read news, we have team meetings, we do performance reviewswe do all of the things that people who don’t code also do, she says. The glaring question around agents like Codex and Claude Code is how they will affect human jobs, especially those of younger engineers. OpenAI wants its agent to behave like a talented assistant engineer but stops short of saying it will replace people. Instead, Sottiaux sees coding agents as a way to expand how teams approach problems and develop new ideas, particularly when less experienced engineers use them to experiment and push beyond conventional approaches. “And then they come up with completely new ideas that you might not have if you anchor too much on your decades of experience, he says.
Category:
E-Commerce
When the email pinged in my inbox, I didnt even bother to open it immediately. I already knew what it was. One glance at the subject line told me everything. After enough time on the job hunt, you develop a sixth sense for HR language. The preview textThank you for taking the timesaid it all. Its the standard soft intro to bad news: Your application was amazing . . . but not amazing enough. The blow softens once youve received a few of these. But the emotions that follow resemble the five stages of grief: denial, anger, bargaining, depression, and eventually, acceptance. I ran the gamut of these feels when I got my latest rejection for a role that seemed promising all the way through the final interview. Heres how I felt and acted after I opened that message and faced reality. Denial Nah, this can’t be right. I refresh my inbox three times, as if the letters in the message will magically rearrange themselves into a sequence that reveals a start date. Could it be a system glitch? Maybe they sent this to the wrong candidate? (Believe it or not, its happened to me before.) I mean, I was perfect for this role. Remember in the final interview when I gave that answer about cross-functional collaboration that made the hiring manager nod so hard I thought she had that new J. Cole playing in her AirPods? I draft a response. Thank you for your consideration. However, I believe there may have been an error . . . I let it sit in my drafts folder for exactly 11 minutes before deleting it. Even my delusions have limits. But I do check LinkedIn to see if they’ve posted the position again. They haven’t. Which means they hired someone. Which means this is real. Which leads me directly to . . . Anger I’m in my feelings now. Who did they hire? I need to know immediately. I’m on LinkedIn doing forensics like I’m on The First 48. I filter the companys employees by most recent hires. There he is. Brayden. Of course it’s a Brayden. His profile says he thrives in ambiguous environments and has experience with stakeholder management. My profile says the exact same thing but with better action verbs. Ugh. Bargaining Okay, let me think about this objectively. What could I have done differently? Maybe I shouldn’t have mentioned I needed to check the start date because of a vacation I had already booked. Maybe that made me seem uncommitted. Or maybe I should’ve asked more questions at the enddid I seem too confident? Not confident enough? Maybe I talked too much . . . or too little. Should I have laughed at the hiring managers joke about getting her ducks in a row? It wasn’t funny, but maybe that was the test. I consider emailing the recruiter to ask for feedback. Just a friendly note. Hey! Would love to learn what I could improve for next time :) The smiley face is crucial. Makes me seem coachable and not at all dead inside. I type it out. I don’t send it. I know what they’d say anyway: We had many qualified candidates. Translation: Brayden’s uncle plays golf with the CEO. Depression It’s been three days since the rejection. I’m still thinking about it. I’ve applied to 16 other jobs since then. Each one feels like I’m rolling up a resume, stuffing it into a Dos Equis bottle, and chucking it into the ocean. My Easy Apply count on LinkedIn is getting embarrassing. I’m tailoring cover letters for positions I’m overqualified for, underqualified for, and in some cases, not even sure what the job actually is. Customer Success Champion could mean literally anything. I think about Brayden again. Brayden’s probably in orientation right now, getting his company laptop, meeting the team, hearing about the unlimited PTO that no one actually takes. Brayden’s probably not wondering if his name sounded too ethnic on the application. Brayden’s probably not calculating whether the commute is worth it while also knowing he won’t get the offer anyway. Brayden’s just . . . winning. I eat leftover jerk chicken at 11 a.m. and consider whether this is rock bottom or if rock bottom is a few more rejection emails away. Acceptance (sort of) Here’s what I know: This isn’t personal, even though it feels personal. Corporate America isnt rigged. It just tends to work out beautifully for guys named Brayden. That company wasn’t the one. Maybe the role wasn’t even that good. The Glassdoor reviews mentioned fast-paced environment, which is code for no work-life balance anyway. I update my resume again. Not because I think it’ll make a difference, but because I need to feel like I’m doing something. I tweak one bullet point. I remove an unnecessary comma. I save it as Resume_FINAL_v3_ACTUAL_FINAL_Feb2026.pdf knowing damn well there will be a v4. And then I do what I always do: I apply to another job. Because theres only one thing worse than getting rejection emails, and thats not getting any emails at all.
Category:
E-Commerce
Hello again, and thank you, as always, for spending time with Fast Companys Plugged In. In a remarkably influential 2011 Wall Street Journal op-ed, Netscape and Andreessen Horowitz cofounder Marc Andreessen declared that software was eating the world. From entertainment to commerce to transportation, he argued, startups that were about code at their core were disrupting many of the worlds most deeply entrenched businesses. That was just the beginning, he warned: Companies in every industry need to assume that a software revolution is coming. Fifteen years later, we know that some of the disruptors Andreessen citedsuch as Zynga, Groupon, and Skype (RIP)did not, in fact, eat the world. His larger point, however, played out much as he predicted. Software really does run everything these days. And many of its purveyors are among the most successful companies in the world. Recently, however, Wall Street has been spooked by the possibility of another sea change in the making: AI might be on the verge of eating software. The sudden leap forward in the capability of software-writing LLM tools such as Anthropics Claude Code has investors worried that the corporate behemoths presently making tidy profits by selling subscription-based softwareparticularly for enterprise customersmight find themselves unable to compete with apps coded by AI for very little cost. This theoretical collapse of the software industry is known as The SaaSpocalypse, a name I hate but cant quite avoid acknowledging. (I promise not to bring it up again.) Its reflected in the stock performance of such seemingly robust companies as Workday (down 35% year to date), Adobe (-26%), Salesforce (-25%), Autodesk (-21%), and Figma (-19%). On February 23, after Anthropic published a blog post touting Claudes ability to modernize software written in the 66-year-old COBOL programming language, IBMCOBOLs kingpin for most of that timesaw its biggest one-day stock drop in more than a quarter century. Investors are right to expect that AI will radically change software as a business in the coming years. The evidence is already here, in the form of developments such as Blockthe parent company of Squareannouncing on February 26 that its terminating 40% of its 10,000 employees. Explaining the brutal reduction, CEO Jack Dorsey contended that AI will allow a smaller team to accomplish more and do it faster, and said he was getting ahead of an inexorable industry-wide trend. What happens next remains to be seen, but Block will surely never be the same. Still, Wall Streets apparent belief that AI spells bad news for todays software titans is premature, and possibly just misguided, period. Its certainly heavy on vibes rather than hard data: Mondays dip in the S&P 500 apparently stemmed in part from a dystopian imaginary June 2028 memo published by Citrini Research. Laying out a sweeping nightmare involving AI crushing the U.S. economy, it name-checked specific companies such as DoorDash and Zendesk as being incapable of competing with AI-infused apps and agents. Well, maybe, though even the documents authors admitted they were certain some of these scenarios wont materialize. In a little over two years, it will be possible to assess what Citrini got wrong and right. For now, it remains equally possible to imagine futures in which 2026s software-based kingpins arent mowed down by AI, even if the technologys coding chops will continue to improve indefinitely rather than hitting a wall. For one thing, the software business isnt solely about writing software. It requires selling itsometimes in the form of hefty annual contractsand supporting it when things go wrong. It will be difficult for AI (or even most AI-savvy startups) to take on these tasks outside of the human-powered infrastructure that major software companies have built, often over decades. In Sun Microsystems cofounder Scott McNealys memorable phrase, enterprise customers like having one throat to chokesomeone with the bottom-line responsibility of making them happy. They wouldnt get that by vibe-coding their own in-house replacements for major apps, or buying them from a tiny company offering look-alike equivalents. Instead, they have a powerful incentive to keep doing business with companies that have already shown an ability to deliver. People who use AI to write their own apps might even develop a newfound appreciation for all the ways software suppliers make their lives easier. For instance, last April I wrote about the note-taking app Id vibe-coded for my own use, and said Id put it together in a week. What I didnt know at the time was that Id spend the next 11 months fiddling around with new features, squashing bugs, and stressing over the fact that Inot Apple, Google, or Notionbear responsibility for the apps security and data integrity. Id do it all over again, but because its been great, mind-expanding fun, not because its saved me money or time. Its far too early to conclude that existing software giants wont use AI to grow even more dominant. After all, they have considerable resources to throw at that challenge, and deep knowledge of the industries they serve. AI could be a potent accelerant to their growth, or just a way to slash costs by reducing human headcount. But theres little evidence its on the cusp of figuring out how to build and market products humans will find compelling without plenty of guidance. Even as the technology puts pressure on software companiessay, by introducing enough competition that its tougher to endlessly raise pricesthey might be intrepid enough to find a new path forward. IBM, for example, isnt short on AI savvy of its own; if the company cant find a way to make money from customers wanting to modernize COBOL-based platforms, its IBMs own fault, not Anthropics. Yes, history is full of sobering case studies of once-mighty software companies that gotoverwhelmed by technological change. In the 1990s, for example, the PCs shift from the text-based DOS to the graphical interface of Windows was ruinous to big names such as Lotus, WordPerfect, and Ashton-Tate, none of which bet big enough on Windows early enough. Their miscalculation was unquestionably Microsoft Offices gain. But it doesnt always pan out that way. In the following decade, Office faced a similar threat as productivity migrated to internet-based tools. When Google launched products such as Docs and Sheets, stuffed them with innovative features, and offered them for free, observers thought that might be terrible news for Microsoft. Not so: The company reacted skillfully enough that Microsoft 365, as it calls Office in its current form, is bigger than ever, to the tune of $95 billion in revenue last year. In Silicon Valley, it has become fashionable to tell workers that the only way to remain relevant is to embrace AI rather than fear it. As Nvidia CEO Jensen Huang puts it, Youre not going to lose your job to an AI, but youre going to lose your job to someone who uses AI. The same principle applies to todays software companies. Theyre not going to be killed by AIonly by other companies that are better at seizing the opportunities it offers than they are. Youve been reading Plugged In, Fast Companys weekly tech newsletter from me, global technology editor Harry McCracken. If a friend or colleague forwarded this edition to youor if you’re reading it on fastcompany.comyou can check out previous issues and sign up to get it yourself every Friday morning. I love hearing from you: Ping me at hmccracken@fastcompany.com with your feedback and ideas for future newsletters. I’m also on Bluesky, Mastodon, and Threads, and you can follow Plugged In on Flipboard. More top tech stories from Fast Company If technology could bring traffic fatalities down to nearly zero, why not embrace it?What the elevator can teach us about self-driving cars. Read More Anthropic’s autonomous weapons stance could prove out of step with modern warThe Pentagon is demanding that the AI company remove the safety guardrails from its AI models to allow all lawful uses. Read More Is Apple about to debut a new iPhone camera feature?What is ‘variable aperture’ and why you should care. Read More AI can write now. What happens to reporters?If bots can reliably draft copy, ‘something big’ might be happening to the job of a journalist. Read More Apple killed Dark Sky. Now its creators are trying again with a new weather appAcme Weather brings back the team behind the cult-favorite forecast app, with new features designed to show uncertainty. Read More 15 incredibly useful things you didn’t know NotebookLM could doFrom managing meetings to maintaining your car, Google’s Gemini-powered research tool can provide all sorts of eye-opening revelations. Read More
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||