Every B2B company hits the same inflection point — enterprise customers show up and they need SSO, directory sync, audit logs, and role-based access before they'll move forward. Most teams lose months building that infrastructure. It doesn't have to be that way.With WorkOS you get all of it. One platform for auth, identity, and security. Infrastructure for teams that ship fast and stay fast.OpenAI, Anthropic, Cursor, and Perplexity already chose WorkOS over building it themselves.Build faster with WorkOS →

Google's new Gemini 3.1 Pro is now rolling out in preview. The model's benchmark results show mostly modest improvements. The updated model is now available in AI Studio and the Antigravity IDE. Enterprise users will see the model in Vertex AI and Gemini Enterprise, and regular users can use it in both the Gemini app and NotebookLM. The API costs for developers have not changed, nor has the context window.

Amazon is now the biggest global company by revenue. Walmart was previously the leader for more than a decade. Jeff Bezos had carefully studied Walmart's founder, Sam Walton, and embedded many of his business strategies while building Amazon. Amazon's revenue increased at almost 10 times the pace of Walmart's over the past decade. Amazon is the biggest online retailer, attracting 2.7 billion visits to its websites and mobile apps each month, while Walmart is the biggest physical retailer in the world, with more than 10,000 stores and shopping clubs globally.

Scaling laws have defined tech in the last five years. Large language models have demonstrated consistent returns to scaling. However, these 'laws' apply to very few other fields. Little work has been done to systematically understand, in theory or in general, what causes scaling laws or determines their slope and robustness. This article looks at what causes scaling laws, what their preconditions are, when they stop, why they have worked in some fields but not others, and more.

Almost everyone agrees that AI policy should be evidence-based. However, science is often messy and full of caveats, which doesn't fit well with policymakers' demands for clean answers. Many AI benchmarks have saturated and can no longer give a clear 'no' on risks. Experiments show how much we still have to learn. We need to spend less time proving that today's AIs are safe and more time figuring out how to tell if future AIs are dangerous.

Traffic skyrocketing? If you're on Microsoft Azure, there's no reason to panic. When user demand spikes, Azure adjusts capacity in the background. Performance stays steady with no manual intervention, no late-night firefighting, and no scrambling to explain why the site went down during the big launch. Build without limits with Azure

The winning formula in early 2026 is structured context, tiered rigor, smaller teams with higher leverage, and relentless measurement of downstream effects. The teams that adapt AI tools without the underlying taste and discipline are discovering that AI just makes their existing problems bigger. Teams with strong engineering cultures, robust CI/CD, clear architectural standards, and effective review processes see AI compound their advantages. This is widening the gap between elite and average teams.

AI should be treated as an amplifier of human capacity rather than a replacement. AI can still act autonomously with specific tasks, but in a way that is an extension of human decision-making and context. This allows humans to do dramatically more work, more sustainably, with less injury and fatigue. The AI industry's dream of having a fully autonomous AI employee is seductive, but it will likely set us up for disappointment.

OpenAI's technology isn't unique. While it has a big userbase, there is limited engagement and stickiness, and no network effects. Its competitors have matched its technology and are leveraging their product distribution. Executing better than everyone else is certainly an aspiration, but it isn't a strategy.

ByteDance is hiring nearly 100 employees for its AI division in the US. The company's AI team, called Seed, was established in 2023 and has labs across the US, Singapore, and China. The open roles include various job responsibilities such as producing international data, advancing ByteDance's text, image, and video generation tools, AI research, and building models for drug discovery and design. The company's recent AI launches have thrust it into the spotlight in the US. There has been some controversy over the data used to train its models.

Tech stack complexity is out of control, and the last thing you need is a bigger ‘stack'. Microsoft Azure gives you a unified view of applications, cloud resources, and AI—so you can ship more software with less open tabs. Simplify with Azure

Software that can be generated in a weekend is disposable inventory, not something you can build a startup around.

Nine observations from building with AI agents over the last year.

Meta is separating its Quest VR platform from its Worlds platform, shifting the focus of Worlds to be almost exclusively mobile.

The most value has always historically accrued to the companies that did the hardest things.

There were around 1,500 federal trade secrets cases filed in the US last year, up 20% from the previous year.

The productivity gains from AI are real, but it won't be catastrophic, as it will just devalue mechanical skills while elevating judgment skills.

As AI evolves at an unprecedented pace, measuring intelligence requires more than a few AI research labs alone – it requires the imagination and collective expertise of the global community. That's why Kaggle launched Community Benchmarks, a new way for anyone to build, design, run and share custom benchmarks for evaluating AI models at no cost.

Community Benchmarks enable you to build evaluations easily. With free access to state-of-the-art AI models (subject to quota), support for multimodal inputs/outputs, tool use, and an easy-to-use interface, you can start building right away!→ Learn more→ Watch the tutorial → Get started

Capital One Financial has agreed to buy Brex for $5.15 billion in cash and stock. Brex launched in 2017 to cater to startups that ran into difficulties getting corporate cards from traditional players. It is one of the youngest US startups to join the billion-dollar valuation club. Capital One expects to close the deal in the second quarter.

TikTok has formed a joint venture to keep its platform operating in the US. Adam Presser, who has worked at TikTok for almost four years and was previously a senior executive at Warner Bros., will be the CEO of TikTok USDS Joint Venture. TikTok CEO Shou Chew will be a director. The joint venture will operate under defined safeguards that protect national security.

Tesla Robotaxi riders in Austin, Texas, are now able to hail a Model Y Robotaxi without anyone in the vehicle. The company has removed the human Safety Monitors that have been present in its robotaxis since launch last June. The change has not rolled out to the entire fleet yet. Videos of riders experiencing the driverless service are available in the article.

Elon Musk says he expects Tesla to begin selling humanoid robots to the public by the end of 2027. Tesla's Optimus robots are already performing simple tasks in the company's factories. There are still profound technological challenges remaining for humanoid robots, including basic issues like hand function. Musk has warned that initial production for Optimus and the Cybercab will be agonizingly slow, but eventually will end up being insanely fast.

There is a lot of power in composing CLIs. With some creativity, it's possible for an individual to maintain CLIs instead of an MCP server. This post discusses how to expose SaaS functionality to agents without MCP. It provides examples of how to do so by giving an agent access to Google Docs and to Google Groups.

GitHub Copilot SDK allows developers to embed the same agentic core that powers GitHub Copilot CLI into any application. It provides programmatic access to the same production-tested execution loop, so developers don't have to manage context across turns, orchestrate tools and commands, route between models, integrate MCP servers, or think through permissions, safety boundaries, and failure modes. The SDK also provides support for multiple AI models, custom tool definitions, GitHub authentication, and real-time streaming. It is now available in technical preview.

Becca Selah joined Amazon in April of 2019, six months after becoming a mother. This post takes a look at her time at the company. It covers Amazon's levels, work/life balance, vacations, vesting, and more. Selah left a year and two days into her tenure with an unceremonious exit.

Netflix is facing increasing skepticism from Wall Street due to its intention to acquire Warner Bros. The decision is a stark departure from Netflix's build-not-buy philosophy. This article contains an interview with Netflix co-CEO Greg Peters where he discusses why engagement is an important metric and how live events play into that, the Warner Bros. acquisition, why Peters is convinced it is a home run, and the company's arguments as to why regulators should allow the acquisition. The interview is available as a podcast.

Microsoft's Windows, Teams, M365, and other divisions have been told to install Claude Code for internal testing alongside Copilot.

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more.

Barret Zoph's firing from Mira Murati's Thinking Machines Lab has resulted in several employees leaving for OpenAI and other companies.

Apple's hardware chief John Ternus' role now includes design work, a role that has long been entrusted to a senior leader.

a16z's State of Markets slide deck shares what the firm sees in the private and public markets, what it thinks the state of AI looks like, and why today's winners are just built different.

Epic helped Google market Android, and in return, Google helped market Fortnite.

The job cuts are part of the company's broader goal of cutting about 10% of its corporate workforce.

Nebius Token Factory is built for teams running open-source LLMs in real products. We deliver managed inference with explicit control over execution paths: speculative decoding, cache-aware routing, and post-training tuned to real traffic. Predictable tail latency, stable cost, and production-grade systems for AI that actually ships.

Build inference that holds at scale →

Apple is reportedly working on a small, wearable AI pin equipped with multiple cameras, a speaker, and microphones. It will likely run the new Siri chatbot the company plans to unveil in iOS 27. Apple wants the final version of the pin to be about the same size as an AirTag but slightly thicker. The AI pin could be released as soon as 2027, but development is still in the early stages, and the product could be canceled.

Apple plans to revamp Siri later this year by turning it into an AI chatbot. The chatbot will be embedded deeply into the company's operating systems and replace the current Siri interface. The feature is a central piece of Apple's AI strategy. Apple has designed the chatbot so that its underlying models can be swapped out over time. The company will initially use Google-powered systems.

Blue Origin plans to deploy 5,408 satellites into space to launch a communications network to take on SpaceX. Its TeraWave network will be targeted at enterprise, data center, and government users. It will provide data speeds of up to 6 terabits per second. Blue Origin expects to start deploying its constellation in the fourth quarter of next year.

Last February, a team of doctors and scientists treated a baby with severe CPS1 deficiency, a condition caused by a rare single-gene mutation that causes patients' blood to be flooded with ammonia that the liver can't clear. The doctors sequenced the baby's DNA, designed a custom gene editor to fix the typo in his CPS1 gene, manufactured it on demand, and infused it into his bloodstream. The process took about six months and cost less than $1 million. Today, the baby is thriving and seemingly cured.

Unblocked is the AI code review that surfaces real issues and meaningful feedback instead of flooding your PRs with stylistic nitpicks and low-value comments.

“Unblocked made me reconsider my AI fatigue. Finally, a tool that surfaces context only someone with a full view of the codebase could provide.” - Senior developer, Clio

Try now for free

Evaluating technical candidates is becoming harder as AI capabilities improve. Each new model forces companies to redesign tests. This post describes one company's original take-home design, how each Claude model defeated it, and the increasingly unusual approaches the company has had to take to ensure its tests stay ahead of top models' capabilities. As work evolves alongside models, companies need to develop increasingly creative ways to find strong engineers.

pandas 3.0.0 includes some new features, bug fixes, and performance improvements, as well as possible breaking changes. It removes a functionality that was deprecated in previous releases. Users are recommended to first upgrade to pandas 2.3 and to ensure their code is working without warnings before upgrading to pandas 3.0. A link to a list of all the changes in the release is available.

X is struggling with two incompatible goals: optimizing attention and time spent on the platform, and tight creator-follow linkage. So far, X has opted to chase KPIs like user time spent on the platform. This has resulted in creators unable to consistently reach their audience, consumers not hearing from creators they want to follow, and filter bubbles like Crypto Twitter. Either creators and users are tightly linked, or the platform algorithmically serves users an array of compelling content - you can't have both.

The rush to build AI data centers prompted Elon Musk to propose putting them in space. Musk has become obsessed with the idea of SpaceX being the first to do it, but such a feat would be hard to attempt and would require the billions of dollars in capital an initial public offering could deliver. A SpaceX IPO could also help xAI catch up to rivals. SpaceX is expected to collect banks to lead the stock offering soon, with the IPO expected to be completed by July.

Cybercab production is scheduled to start in less than 100 days.

YouTube Music is getting more discovery features, and YouTube TV is getting 20 genre-specific plans as well as a fully customizable multiview.

Meta is reportedly working on a large language model and an AI image and video model.

AI is a significant productivity tool comparable to previous technological shifts, but it's boring.

Anthropic trains Claude using a specific set of ethical principles (detailed in its 'Constitution') rather than human feedback.

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more.

The company now has fewer than 15 people left in engineering.

Developer tool sprawl is real. You've got one tab for logs, another for metrics, half a dozen for cloud services, and somewhere in there is the code you're actually trying to ship.

🫣 Stop playing hide-and-seek with your tools. Microsoft Azure brings together everything you need to build and run software in a single workspace.

Goodbye, hunting for the right dashboard. Hello, one unified view—and a faster path to innovation.

Simplify your stack with Azure →

X has open sourced its algorithm, giving the world a peek behind the company's algorithmic curtain. The site's algorithm considers engagement history and surveys recent in-network posts when finding content for individual users. It conducts a machine-learning-based analysis of 'out of network' posts that it believes the user might also find appealing. The algorithm then filters out certain kinds of posts, as well as content deemed too violent or spam-like, and ranks the content based on what it thinks users will find most appealing.

Elon Musk's lawsuit accusing OpenAI and its CEO Sam Altman of abandoning the nonprofit mission has unveiled over 200 documents that offer an inside look into the relationship between OpenAI and Microsoft. Musk, an OpenAI co-founder, is seeking up to $134 billion in damages, with a jury trial scheduled for this Spring. The files reveal how Microsoft pursued, rebuffed, and backed OpenAI at various moments over the past decade. The previously undisclosed emails, messages, slide decks, reports, and deposition transcripts show how Microsoft ultimately shaped the course of the lab that launched the generative AI era.

In a Phase 2b trial, Moderna and Merck's cancer vaccine reduced the risk of relapse or death in melanoma patients by 49% when used in combination with Merck's immunotherapy Keytruda, compared to when using Keytruda alone. The results illustrate mRNA's potential in cancer care. The treatment continues to demonstrate sustained and clinically meaningful improvement. The companies are currently running eight Phase 2 and Phase 3 clinical trials across multiple tumor types.

The International Space Station is due to be decommissioned in less than five years, but NASA has yet to formally publish rules and requirements for a potential replacement. There are currently four main contenders for NASA's commercial space station program: Voyager Technologies, Axiom Space, Blue Origin, and Vast Space. NASA is expected to select one or two of these companies for contracts later this year. This article contains an interview with Vast's chief executive, Max Haot. Vast is currently building a smaller interim space station capable of short-duration stays called Haven-1.

Every new monitoring tool, service, or "quick integration" adds another dashboard to the pile—and before you know it, you're spending more time switching between tools than actually building. With Microsoft Azure, your entire stack lives in one workspace: infrastructure, data, AI, and more. See how Azure brings it all together →

Harnesses are constantly improving by iterating on what works well. However, every time someone releases a new behavior, the thought leaders of the internet determine that it has replaced all previous functionality. Developers should try both skills and MCP to learn the fundamentals of how the technology works and what that means for managing the context window. How the technologies shift in the future depends entirely on the workflows they help people solve.

Skills are reusable capabilities for AI agents. This site contains an open agent skills ecosystem with a collection of skills anyone can install with a single command to enhance their agents with access to procedural knowledge. The site contains thousands of skills. The index is searchable and shows how many times each skill has been installed.

People think software engineers are addicted to being useful, driven by an addiction to solving puzzles, or drawn by power and money. What actually motivates software engineers is often more of an internal compulsion. It's worth figuring out what really motivates you so that you can harness that compulsion most effectively.

The concept of 'Scalable Agency', where a developer could theoretically spin up thousands of agents to complete tasks faster, relies on a flawed assumption that software engineering is an embarrassingly parallel task. Context loading is not the same as common knowledge, and reading tokens is not the same as understanding the causal chain of changes across a system. Adding manpower to a late software project makes it later. Multi-agent systems do not escape this law, they simply hit the wall of state-space explosion faster than humans do.

syntux is a generative UI library that is consistent and flexible.

Just like when working with clay, when software doesn't work, you throw it back, refine it, and spin again until it's right.

AI agents help facilitate better match quality through centralized mechanisms in 'thin' markets.

Prediction markets are creating a systematic wealth transfer, where impulsive Takers pay a structural premium for affirmative outcomes while Makers capture an 'Optimism Tax' simply by selling into this biased flow.

pipenet is an alternative to localtunnel that bundles client and server to host self-hosted tunnel infrastructure for full control over security, domains, and availability.

Netflix originally planned to buy each Warner Bros. share with $23.25 in cash and $4.50 in Netflix stock, but it has amended the deal to sweeten the offer for shareholders.

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more.

Bugs sneak out when less than 80% of user flows are tested before shipping. However, getting that kind of coverage (and staying there) is hard and pricey for any team.

QA Wolf's AI-native solution provides high-volume, high-speed test coverage for web and mobile apps, reducing your organization's QA cycle to minutes.

They can get you:

• 80% automated E2E test coverage in weeks—not years
• Unlimited parallel test runs
• 24-hour maintenance and on-demand test creation
• Zero flakes, guaranteed

The benefit? No more manual E2E testing. No more slow QA cycles. No more bugs reaching production.

With QA Wolf, Drata's team of engineers achieved 4x more test cases and 86% faster QA cycles.

Get Started For Free

TikTok's new app, PineDrama, offers access to microdramas, bite-sized TV shows that can be watched in a series of one-minute episodes. It features a variety of genres, including thriller, romance, and family. The microdrama industry is projected to bring in $26 billion in annual revenue by 2030. PineDrama will compete with existing microdrama platforms like ReelShort and DramaBox.

OpenAI is on track to unveil its first hardware product in the second half of 2026. Jony Ive's 'io' team at OpenAI recently hired Janum Trivedi, who previously worked on SpringBoard at Apple. Trivedi also built Split View, Multitasking Drag & Drop, iPad Pointer Gestures, and more for iPadOS 15.

Matic's robots have cleaned over 50 million square feet across real homes since launch. The company was designed from the ground up to iterate quickly. It shipped early, observed behavior, learned from real homes, and refined constantly. This article tells the story of Matic's founders and how they built the company.

NASA rolled out the Artemis II rocket to its launch pad over the weekend. The Artemis II mission will set several notable spaceflight records. The astronauts will travel further from the Earth than any human in history, flying to the far side of the Moon and back. The mission is on track for liftoff as soon as next month.

Connect your AI agents to apps and data more securely, give users control over the actions AI agents can perform and the data they can access, and enable human confirmation for critical agent actions.

Start building with Auth0 today.

Claude Code 2.0 has a flexible and robust harness with an evolved UX and fewer bugs. The RLHF Anthropic did on Opus 4.5 completely changed the equation. Users no longer need to review code or instruct the model at the level of files or functions: they can just test behaviors instead. Claude Code can be used by developers who never plan on learning how to code and just care about outputs.

Interesting software development jobs aren't likely to disappear. The jobs that are dependent on the fundamentals of software aren't going to stop being dependent on the fundamentals of software. More LLM usage will mean more stress on the tools and applications that rely on the fundamentals of software. The most conservative way to build a career as a software developer is to be practical and effective at problem-solving and to better understand over time how the tools you use work so that you can make better decisions.

The New York Stock Exchange is building a new digital trading platform using blockchain technology to allow for around-the-clock trading of tokenized stocks and exchange-traded funds. The platform will launch later this year, pending regulatory approval. It will allow for new types of investor accessibility and create new opportunities for retail to participate in stablecoin-funded markets. The digital infrastructure will allow trades to be funded and settled in real-time without the current one-day delay.

Nearly $12 billion was traded on Kalshi and Polymarket in December, up more than 400% from a year earlier. The rise of prediction markets has created new risks and raised fears about the spread of gambling. There are also concerns about market manipulation and insider trading. Proponents of prediction markets claim that the platforms are fundamentally different from gambling and that they offer a valuable new source of information by allowing people to bet on world events.

75% of leaders believe that AI is overly focused on individual productivity, creating siloes and hurting collaboration. What's the alternative? Get a free copy of the study from Miro

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more.

AI isn't competent enough yet to program in COBOL, and keeping code compliant remains a tough challenge.

No amount of brilliant UX, go-to-market strategy, or engineering can make customers adopt a product whose core AI task doesn't solve their job to be done.

Oculus founder Palmer Luckey says Meta was crowding out the ecosystem and that recent studio shutdowns and layoffs will be a good thing for the long-term health of the VR industry.

AI agents are massive slop machines if you turn off your brain and let go immediately.

The change will reduce the chance of customers getting counterfeits from Amazon when buying from a reliable seller.

A recent change removes the blue background around sponsored results.

Over 6 million tech professionals read TLDR including developers, product managers, marketers, designers and executives. Get in front of your target audience with an ad placement just like this one! Learn more about running a test campaign.

OpenAI plans to start testing ads in the US for Free and Go tier users in the coming weeks. Plus, Pro, Business, and Enterprise users will not see ads. The Go tier is a new $8 per month subscription with access to GPT-5.2 Thinking. Ads will not influence the answers that ChatGPT provides and will always be separated and clearly labeled. Conversations will be kept private from advertisers. A screenshot of what the ads will look like is available in the article.

Elon Musk wants OpenAI and Microsoft to pay him damages in the range of $79 billion to $134 billion as he claims OpenAI defrauded him by abandoning its nonprofit roots and partnering with Microsoft. The court filing says Musk is entitled to a chunk of OpenAI's current $500 billion valuation as he donated $38 million in seed money when he helped found the startup in 2015. Musk also plans to seek punitive damages and possibly an injunction. OpenAI claims the lawsuit is baseless and is aimed at harassment.

Sid Sijbrandij, founder of GitLab, got the news that he had cancer on November 18, 2022. He underwent a brutal care regimen that involved surgical removal of his tumorous vertebrae and rounds of radiation and chemotherapy so intense that he required four blood transfusions to stay alive. After his cancer resurfaced in 2024, he decided to go 'founder mode' on his cancer, assembling a team to navigate his care journey. This article details the story of how he took charge of his health.

While the humanoid form factor is likely to stay due to its clear advantages, that doesn't mean that robots will still stay human. Boston Dynamics' Atlas is an interesting look at a very different vision of what a humanoid robot can and should be. Robots don't have to be limited to what evolution resulted in. They can be engineered to have capabilities that far exceed humans.

Over 70% of orgs are using AI agents, but only 11% of agentic use cases made it to prod in 2025. In this webinar, Camunda's CTO joins automation experts to cover the proven controls that help AI agents progress. Secure your spot to learn how to move AI agents from pilot to prod.

This article details how a developer created an image-to-ASCII renderer. The resulting renderer converts 3D models and still images into ASCII. The renderer is interactive, and you can move the 3D models around by dragging. The post is detailed and contains many interactive examples explaining the work.

Files represent personal creations, so they should live somewhere users control. File systems allow data to outlive the software used to create it. Other apps are able to open the same files to read or write data, and it's up to the users' choice on what app to use. This is analogous to how the AT protocol (the protocol behind Bluesky and other social apps) works.

Agents have improved over the last year due to structures that provide them with automated feedback and allow them to work on longer-horizon tasks. This back pressure allows agents to identify mistakes, and the feedback helps them align on tasks for much longer. Developers should think about how to build back pressure into their workflows. Afterwards, they can loop agents until they have stamped out all of the inconsistencies and issues.

We're hiring the first engineer on TLDR's new Applied AI team. You'll build production AI agents and design modular, composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more.

Prediction markets let people wager on basically anything. These sites have recently boomed in popularity and are creeping into the mainstream. They are transforming news into a form of gambling. It is unclear whether these sites are meaningful predictors of anything, but the markets are manipulable.

Granola is AI that transcribes your meetings and enhances your notes - so you can give each call your full attention. The best part? There's no meeting bot. Try Granola free

Increased productivity means there's more that you could be doing.

Most companies are bad at ops, and this is becoming clear as AI adds more chaos to systems already defined by chaos.

Developers should define the rules LLMs operate under instead of rejecting and blocking them.

This post discusses Brex's culture, how it facilitated an aggressive shift towards a culture of AI fluency, and how Brex modernized its tech stack and applied AI agents to streamline its business.

Tesla plans to iterate rapidly, aiming for a nine-month design cycle.

jQuery 4.0.0 is the first major version release in almost 10 years and includes some breaking changes.