Why pick one?

With Marqeta, launch payments experiences without choosing between innovation and scale.

Marqeta combines the scale and reliability of proven payments infrastructure with the flexibility and innovation of a modern platform—so you can move faster, reduce risk, and grow with confidence.

From optimizing spend and cash flow to building seamless, rewards-driven customer experiences, Marqeta's solutions have you covered. When it comes to issuing and processing payments, don't sacrifice. Choose a payments provider who delivers more. See what's possible with Marqeta >

Learn more >

Apple's new Mini Apps Partner program gives Apple a 15% cut of in-app purchases for mini apps and games within larger 'super apps'. Mini app creators have been using loopholes to avoid Apple's in-app payments framework. The new program should help lock those software creators into Apple's preferred approach and guarantee more revenue for the company. Mini apps or games must be released by a party not directly or indirectly controlled by the developer of the dominant app to qualify for the new program.

Blue Origin nailed the landing of its New Glenn rocket booster yesterday. It is the second company to successfully accomplish the feat. New Glenn's two successful launches in a row could win it a sizable slice of the business of sending stuff to space. The New Glenn booster is considerably bigger and heavier than SpaceX's Falcon 9 booster.

Trying to predict the future is an excellent feedback mechanism for developing a world model in areas you care about. You can review your predictions to see what you got wrong and why. This site presents a timeline for AI development that predicts that a continuously improving AI that learns from interactions in the real world, with the potential to become more expert at most tasks than even humans with 40 years of experience, won't be developed until after 2045. Winning robotics is a national security issue - technology that gives its owner access to users and resources at scale is a threat. There will be a diversity of robot form factors in the future.

The Chan Zuckerberg Initiative's Biohub has a billion dollars a year to make several long-horizon bets on basic science. Its aim is to change our understanding of immunological science and open up the door to engineering human health, simulating immune therapies, reprogramming dysfunctional cells, and preventing diseases before they arise. It has opened up three models on its virtual cells platform: VariantFormer, a model that directly translates personal genetic variations into tissue-specific gene activity patterns; CryoLens, a large-scale model for cryoET that provides unsupervised structural similarity analysis; and scLDM, a model that can generate realistic single-cell data in silico at unprecedented fidelity. A transcript of an interview with Mark Zuckerberg and Priscilla Chan, where they discuss the Biohub secret master plan, is available in the article.

We love this practical guide from You.com that walks through a proven framework to identify, prioritize, and document high-value AI opportunities. Explore where AI can drive the most impact in your organization, both internally and externally. Learn more with this AI Use Case Discovery Guide.

OpenAI has released GPT-5.1 on its API platform. The model can dynamically adapt how much time it spends thinking based on the complexity of tasks, making it significantly faster and more token-efficient on simpler everyday tasks. It features a 'no reasoning' mode to respond faster on tasks that don't require deep thinking. OpenAI has extended prompt caching for up to 24 hours to drive faster responses for follow-up questions at a lower cost. Priority Processing customers will experience notably faster performance with GPT-5.1 over GPT-5.

GenUI (Generative UI SDK for Flutter) is a Flutter library that lets developers easily add interactive generative UI to their applications. It is designed to help developers replace static walls of text from large language models with dynamic and interactive graphical UIs. It can turn conversations and agent interactions into rich, intuitive experiences. State changes in the UI are fed back to the agent, creating a high-bandwidth interaction loop.

The Tesla engineers behind Autopilot, robotaxis, and the Optimus robot have been told to brace for the 'hardest year of their lives' in 2026. The company aims next year to turn a decade of autonomy promises into something commercial, scalable, and visible to investors. Tesla's next era is tied to two products that are nowhere near mass deployment. Elon Musk has previously claimed that Tesla's Optimus robot could become the largest product in history.

The most helpful mentors are usually folks who are only a few years ahead of you. A person who is a decade ahead may be in an objectively more impressive place, but someone nearer to your stage will have experienced an environment that is much closer to your own. This post contains advice for college freshmen seeking mentorship on managing their time and looking for their first job.

TLDR AI is your daily fix of LLMs, GenAI, and deep learning goodness. Same TLDR format. Still free.

Subscribe now.

Discover how Atlassian's Teamwork Collection is helping teams connect work, knowledge, and goals in one intelligent platform. Read the full story to see what this means for teams everywhere.

It's becoming clear that AI startups are a sinkhole for losses.

Treat drag like the threat it is: make it visible and ruthlessly eliminate it.

A look at what it took to bring the Pebble 2 Duo to production and what is coming next at Pebble Tech Co.

Elon Musk hasn't previously opined on the omission of CarPlay from Tesla's vehicles, but he has frequently criticized Apple.

AI agents aren't there yet, and more people are starting to admit it.

Baseline adoption has grown - it is now time to improve it based on real user feedback to have an even greater impact.

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.

Schedule a demo to learn more

Anthropic expects to break even for the first time in 2028. OpenAI's operating losses for 2028 are forecasted to swell to about $74 billion - it doesn't expect to turn a profit until 2030. OpenAI is investing far more into its chips and data centers to turn OpenAI into a multitrillion-dollar tech giant. The strategy requires near-constant fundraising, which could backfire if markets cool on the technology or its near-term profitability.

Apple will delay the launch of the next-generation iPhone Air that was scheduled to be released with the iPhone 18 Pro. The iPhone Air has faced poor sales and manufacturing cuts. The novel look of the device failed to spark interest, despite its substantial design update. Apple is planning a split launch for the iPhone 18 models - the iPhone 18 Pro, iPhone 18 Pro Max, and foldable iPhone are planned for the fall of 2026, while the iPhone 18 and iPhone 18e will be released in early 2027.

NASA's ESCAPADE mission is set to launch tomorrow at 2:50 PM EST on Blue Origin's New Glenn rocket. The mission was delayed last year as its rocket wasn't ready to send it to Mars during its appointed launch window. Mars is now more than 220 million miles away from where it was during last year's launch window - this isn't a big deal, as the mission will follow a 'kidney-bean' shaped trajectory to reach the planet. Such techniques are necessary if humans are to send hundreds of spacecraft to Mars, as it would be difficult to launch only within certain timeframes for direct flights.

Louis Andre, a 27-year-old who grew up in Europe, recently revealed a new company backed by Sam Altman called Episteme. Episteme is an effort to attract the world's top scientists and have them work on a wide range of breakthrough products. It will allow scientists to do away with fundraising pressures and grant writing and spend most of their time on research. The company will also help scientists deal with intellectual property concerns, tax issues, hiring, and other day-to-day support functions to help them turn their ideas into blockbuster products.

Unifying frontend and backend observability delivers measurable gains - less downtime, lower costs, and better-spent engineering hours. Learn how to achieve full-stack observability with a free copy of this eBook from Datadog.

git-rewrite-commits is an AI-powered git commit message writer. It can automatically rewrite entire git commit histories with AI. The tool is perfect for cleaning up commit histories before open-sourcing projects. It can also aid in improving repository maintainability.

Compiler engineers are software engineers who implement programming languages. There is very little information online about how to break into this niche - this post aims to help those interested in the field understand what working in the role is like and how to get a job working with compilers. Compiler engineer roles are relatively rare due to limited demand - once a compiler is built, the work is mostly maintenance and optimization, and most companies don't build their own compilers. The job market can be brutal, but it's possible to stand out even without a preexisting network.

The twenty-first century is starting to build machines and systems that don't need as many humans. Automation is a slow and uneven pressure that makes jobs appear as well as disappear. Right now, the good jobs at the center of the curve are thinning out, and junior entry points are being quietly closed. Graduates are being told to become outliers while competing with systems that learn from everything they do.

Smartphones have become humanity's second cognitive organ, but using the devices harshly disconnects users from the world around them. The only form factor that seems to have a chance to replace the smartphone is AR glasses. Meta Ray-Ban Display glasses are still highly dependent on mobile devices, and they only provide a small display visible to one eye, but they give users a glimpse into what the future may look like. This article provides a detailed review of the device along with the Meta Neural Band that pairs with it. The technology is still very much a first-generation product, but it shows promise.

TLDR AI is your daily fix of LLMs, GenAI, and deep learning goodness. Same TLDR format. Still free.

Subscribe now.

If AI coding feels risky, you're not wrong. But with multi-layer context, you can get agents to respect your architecture, standards, and compliance rules. See how Tabnine bridges the gap

Finding your first user is more of a search process than a persuasion process.

Elon Musk says Tesla will build 10 million Optimus units every year in Texas.

A long reflection on Ruby's past, present, and future.

This post presents a comprehensive analysis of the semiconductor industry that covers what chips are, how they are made, who makes them, fabless vs foundries, China's rise, US reshoring, and investment opportunities.

AI cheating shows how the education system today has no value.

Facebook's Like and Share buttons for third-party websites will be discontinued on February 10.

Warp fuses the terminal and IDE into one place, with AI agents built in. Edit files, review diffs, and ship code, all without leaving the platform that is trusted by over 600k developers and ranks ahead of Claude Code and Gemini CLI on Terminal-Bench.

Ask Warp agents to:

• Debug your Docker build errors
• Summarize user logs from the last 24 hours
• Onboard you to a new part of your codebase

Download Warp for free and get bonus credits for your first week.

Tesla's in-house Rental program lets people rent vehicles for between three to seven days, starting at $60 per day. It includes free Supercharging and free Full Self-Driving for the duration of the rental. There are no limits on mileage or charging, but renters are not allowed to leave the state from which they rented the vehicle. Customers who purchase their own Tesla within seven days of using the Rental service will receive up to a $250 credit towards their purchase.

Apple is working on new satellite features, including a satellite framework for third-party apps, satellite-powered maps, richer messaging capabilities, 'natural usage' improvements, and satellite over 5G. Bringing most of these new features to market will require major upgrades to Globalstar's satellite infrastructure. SpaceX acquiring Globalstar would force Apple to rethink its business model and long-term strategy for satellite services. However, it could also bring opportunities - Apple could collaborate with SpaceX on premium satellite features.

Baidu, Pony AI, and WeRide each have hundreds of robotaxis on the road in China operating commercial paid services without a human safety driver. Their technology and rider experience are generally similar to the driverless taxis operated by Waymo in the US. China's robotaxi fleet is projected to grow to tens of thousands of vehicles by the end of next year. The Chinese companies are testing autonomous vehicles in around half a dozen countries - Waymo is only testing its vehicles in one country outside of the US.

Preventive, a startup backed by OpenAI CEO Sam Altman and his husband, as well as Coinbase CEO Brian Armstrong, is working toward creating genetically engineered human children. Editing gene embryos with the intention of creating babies is banned in the US and many countries, so the company has been searching for places to experiment where the practice is allowed. Preventive is part of a growing number of startups funded by powerful people in Silicon Valley that are pushing the boundaries of fertility and working to commercialize reproductive genetic technologies.

AI spend is rising, but how are you measuring return on investment? We love this guide from You.com, which gives leaders a step-by-step framework to measure, model, and maximize AI impact. Get clear KPIs, four ROI formulas (and when to use each), and a battle-tested LLM prompt for quickly creating your own interactive ROI calculator. Turn “we think” into “we know.”

Download the AI ROI Guide

Valdi is a cross-platform UI framework that compiles TypeScript directly to native views on iOS, Android, and macOS. It delivers native performance without sacrificing developer velocity with automatic view recycling, optimized component rendering, an optimized layout engine, and viewport-aware rendering. Valdi eliminates the traditional compile-test-debug cycle that slows native development with instant hot reload, full VSCode debugging, and a familiar syntax. It integrates easily into existing apps and can be scaled as needed.

Work is much messier today than when roles were somewhat rigid and handoffs were clean. Loops are now tighter, and the fast path from idea to impact crosses multiple disciplines. It is now better to have cross-domain fluency along with product sense and engineering craft. AI doesn't lower the cost of choosing the right thing, shaping it well, and operating it in the wild. Models can accelerate execution, but developers still own intent, architecture, and accountability.

There's a lot of mystique about Cursor. This article takes a look inside the company from the perspective of a team member who has been there for two months. The company's leaders are enthusiastic about establishing a new playbook for company building. The employees are focused on the mission - the company has made them wealthy, but there is zero chatter from employees about getting rich.

China has a million people working on AI 24/7, while the US has maybe 20,000 in Silicon Valley. US export controls are producing precisely the opposite of their intended effects and accelerating China's development. They have triggered a national mobilization. China is predicted to have more AI compute than the rest of the world combined by 2027.

TLDR AI is your daily fix of LLMs, GenAI, and deep learning goodness. Same TLDR format. Still free.

Subscribe now.

The External Secrets Operator is a Kubernetes operator that integrates external secret management systems, reads information from external APIs, and automatically injects the values into a Kubernetes Secret.

China is on course to beat the US back to the Moon with humans.

UnisonDB is a database designed specifically for Edge AI and Edge Computing that enables near-instant fan-out replication across hundreds of nodes while preserving strong consistency and durability.

There are always tradeoffs - recognizing them and making them deliberately is one of the highest value things you can do to shape your career.

Dithering replaces shades with fewer colors strategically placed to maintain the original look.

Mark Zuckerberg and Priscilla Chan will refocus their philanthropy to the intersection of biology and AI to help cure disease.

Considering a Google Developer Premium membership? Google recently expanded the availability of its monthly membership plan so more people can experiment with the full breadth of Google's developer tools, including Gemini Code Assist.

Sign up for a yearly membership to get benefits including:

• A 1-hour consultation with Google Cloud experts.
• $550 in GenAI & Cloud developer credits.
• Unlimited Google Skills.
• 1500 Gemini CLI requests per day.
• 1 Google Cloud certification voucher.

Start building faster with hands-on labs, Gemini CLI, and more exclusive benefits from the Google Developer Program premium membership.

Join today to claim your benefits >

Tesla shareholders have voted to approve a plan to motivate Elon Musk with a bonus of as much as $1 trillion in additional stock. Musk had threatened on social media to leave the company if the measure was rejected. He claimed he wanted a big enough ownership stake in Tesla so that he couldn't be fired, so the 'robot army' he is developing doesn't fall into the wrong hands. The new pay package includes 12 chunks of stock that could give Musk control of as much as 25% of Tesla if he hits a series of milestones in the next decade.

Google's latest generation of Ironwood accelerators will become generally available in the coming weeks. The TPU v7 accelerators are a major leap in performance over prior generations. They offer performance close to Nvidia's Blackwell GPUs, when normalizing floating point perf to the same precision. The TPUs can be scaled into truly enormous compute domains - Google is offering the chips in pods of 256 up to 9,216.

The UAE's Emirate Genome Programme started six years ago with the goal of mapping the genetic code of all of the nation's around one million citizens. The country believes that gene sequencing is both a public-health project and a key part of the country's multitrillion-dollar effort to refashion itself from a petrostate into an AI powerhouse. The project has already collected DNA samples from more than 750,000 Emirates. Efforts to pair AI with genomic data have yet to yield profitable drugs, but many believe they will soon.

'Mind captioning' is a method developed at NTT Communication Science Laboratories in Japan that uses a blend of brain images and AI to generate text descriptions of what people are seeing (or even visualizing) based only on their neural patterns. In tests, the system was able to correctly describe scenes viewed by participants about half the time, even when presented with a hundred possibilities - and even when participants were just visualizing the scenes in their mind. The technology is still far from being a mind-reading device, but it could have serious ethical implications when it becomes more capable.

Employees in large organizations have an average of 112 tools — at a cost of $5,000 per person, per year. This Forrester report presents a framework to consolidate your tech stack strategically, rather than making arbitrary cuts. Look beyond ROI to discover which tools actually drive productivity. Get the report free from Miro

Some concepts are easy to grasp in the abstract, but others you really need to try before you understand them. Regardless of whether or not LLM agents are snake oil or not, it's worth learning how to make one - startups are raising millions building agents at a time when nobody really knows anything yet. It's easy to get started. This post walks readers through how to create an agent.

As large language models become more dominant, new frameworks will find it hard to gain traction because the models won't have enough training data about them. The industry needs to innovate and build new frameworks, libraries, and platform features to push the web forward and create competition. Developers need to have clear strategies to get their work into the LLM training corpus, system prompts, and developer minds. If the industry continues its current focus on maintainability and developer experience, we'll end up in a world where the web is built by LLMs using only what is entrenched in the training data.

This post presents a deconstructive view on how games are designed. It presents twelve aspects of game design to help readers get better at making games of any type. Each of the topics is deep, and there are plenty of links for further reading. The fun of making games is always right outside the edge of what the designers know.

OpenAI's chief financial officer, Sarah Friar, raised concerns after she mentioned the prospect of government aid for the company at a recent conference. Friar has since made a post clarifying her statement, saying that she was making the point that US strength in technology will come from building real industrial capacity, which requires the private sector and government to play their parts. David Sacks, the White House's artificial intelligence and crypto czar, said the federal government has no intention of providing any kind of bailout to AI companies.

61% of customers leave after one bad service moment. Discover how leading brands pair agent empathy with AI efficiency to deliver smarter, more trusted customer experiences. Read the ungated insight

Why pick one? Build a payments solution tailored to your customers' precise needs and to match the scale of your business.Get the best of both worlds with Marqeta.

Meta shows users an estimated 15 billion 'higher risk' scam ads every day.

The new feature will likely be locked to certain regions, so it probably won't be made available to those outside of Europe.

Advice on making Android notifications perfectly reliable from a developer with years of real-world experience building apps with zero notification failure reports.

An easy fix is a pragmatic step forward that delivers value immediately - it can always be refactored later.

The Cybercab, which won't have pedals, a steering wheel, or side mirrors, will be starting production in April at Tesla's factory in Austin, Texas.

Microsoft's new MAI Superintelligence Team, led by Mustafa Suleyman, will put human interests and guardrails first, creating systems that are aligned to human values by default.

Enterprise customers expect more than a login screen. They demand SSO, directory sync, granular permissions, and detailed audit logs, all built to strict compliance standards.WorkOS gives growing teams these enterprise foundations without slowing development:- Modular APIs for authentication and access control- A hosted Admin Portal that simplifies customer setup- Built-in security and compliance features enterprises requireTrusted by OpenAI, Cursor, and Vercel, WorkOS powers auth, identity, and security for AI products across the industry. Your first million MAUs are free.Start building with WorkOS today →

Apple plans to pay Google $1 billion a year for a 1.2 trillion-parameter AI model to help run its Siri voice assistant. The model will run on Apple's own Private Cloud Compute servers, ensuring that user data remains secure. The companies are still finalizing the agreement, which will be an interim solution until Apple's models are powerful enough. The new Siri is on track to launch in Spring.

Blue Origin is getting ready for the second launch of New Glenn, which is scheduled to lift off from Cape Canaveral, Florida, no earlier than 2:51 PM ET on Sunday. The heavy-lift launch vehicle is now in the final stages of preparation for the flight. The mission will include another boost recovery attempt and the launch of NASA's ESCAPADE mission. The success of the launch will determine whether New Glenn could become a viable option for critical NASA science missions and national security contracts.

Quantinuum's Helios machine contains 98 physical qubits, and from those, it can deliver 48 logical error-corrected qubits. The 2:1 ratio is unique - other machines require anything from dozens to hundreds of physical qubits to create one logical qubit. Quantinuum expects to create a large-scale fault-tolerant machine by 2029 with its upcoming Apollo machine, which will have thousands of physical qubits and hundreds of logical ones. Its customers include JPMorgan Chase, which has been using Helios in private preview to run complex algorithms.

More than 10,000 infants in the US are born prematurely each year. Premature birth remains the nation's second-leading cause of infant death, and those who survive it may face crippling lifelong complications. Artificial wombs promise to save more babies, but the technology raises ethical questions that need to be answered before human trials begin. If perfected, the technology could rewrite the limits of viability.

Gartner's annual report looks at 11 popular collaboration tools. Atlassian was named a Leader for its Teamwork Collection — Jira, Confluence, Loom, and Rovo — which connects work, knowledge, and goals in one AI-powered system across every team. Find out how your teams can benefit from this foundation on the Atlassian blog, where you can also get a copy of the report.

There are currently many mature options for compiler targets. This post presents a short survey of some of the popular and interesting options. Compiler targets can include machine code, other high-level languages, virtual machines, and more. Some compiler developers pick unconventional targets for academic curiosity, artistic expression, or to test the boundaries of viable compilation targets.

This post contains a list of markup/configuration languages along with commentary about what's wrong with them. Some specifications are too big and contain too many features, some are no longer maintained, and others are too ambiguous. All languages have trade-offs.

Andy Jassy recently laid out a vision for how he wants Amazon to operate like the world's largest startup. The company has been working hard to flatten its organization, getting rid of bureaucracy in order to move faster and stay competitive. Amazon announced last week it would lay off about 14,000 corporate employees and said more cuts are expected soon. The next big wave of cuts is expected to start in January. Jassy has been searching for Amazon's next opportunity - the company has made big bets on satellite internet, health care, grocery, entertainment, and self-driving vehicles.

Google and Epic have reached a settlement that will provide a framework for long-term global changes to Android app distribution. It still needs to be approved by a judge, but the changes could be put into practice by late this year. Third-party app stores will get a boost, developers will enjoy lower fees, and Google won't drag out the process for years. The agreement isn't intended to undo the jury verdict or sidestep the court's previous order, but is aimed instead at reinforcing the court's intent while eliminating potential delays in realigning the app market.

Ship faster with instant APIs, secure and flexible data, and managed infra, all in one backend. Power any frontend, deploy in minutes. No servers or setup required. Try Free

Apple recently held an event where it detailed how to produce Apple Immersive Video for the Apple Vision Pro.

Large language models write unconstructive, noisy, brittle, and downright bad unit tests.

While the Internet Archive survived its copyright battles, the fight led to more than 500,000 books being removed from its library.

Netflix has sent dozens of requests to agents to add more video podcasts to a platform it plans to launch in early 2026.

Longtime Sequoia investors Alfred Lin and Pat Grady will succeed Roelof Botha, who was asked to step aside after some Sequoia partners raised concerns about his leadership style.

The current generation of AI tools often have 'create and share' viral loops that amplify whatever top of funnel comes in.

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.

Schedule a demo to learn more

Elon Musk spent most of his time at xAI after he left DOGE in May, even holding meetings for Tesla employees at xAI's offices. Musk has balanced his responsibilities between several fast-growing companies for years, but the potential of AI has led him to spend much more of his time at xAI. Tesla will soon announce preliminary results on a giant pay package designed to ensure that Musk focuses on Tesla for years to come. The package would increase his stake in Tesla over a decade from 15% to around 25% if he hits ambitious goals.

Apple is developing an entry-level budget Mac designed for students, businesses, and casual users. The low-cost laptop is currently in active testing at Apple and in early production with overseas suppliers. It will be priced well under $1,000. Apple plans to launch it in the first half of next year. The move is a strategic shift for Apple, which has previously vowed not to chase market share with lower-end offerings.

The cost of electricity on Earth is a problem for large data centers. Google plans to put data centers in space in a dawn-dusk sun-synchronous low-earth orbit to ensure they get almost constant sunlight exposure. It will require solar panels that are up to eight times more efficient to get the job done. The project will only be successful if Google's TPUs can run for at least five years. Google plans to launch a pair of prototype satellites by early 2027.

Amazon is currently laying a new subsea cable across the ocean floor that will connect the US and Europe. It should be up and running in 2028. Dubbed Fastnet, the state-of-the-art fiber-optic cable system will be able to transport over 320 terabits per second. Amazon is investing heavily in the burial process to prevent Fastnet from being cut or damaged.

Meet Nebius Token Factory,the enterprise platform for running open-source models in production. Deploy Qwen, GPT OSS, Kimi, DeepSeek, Llama, and much more on dedicated endpoints with 99.9% uptime, zero data retention, and cost-efficient autoscaling. Optimize, fine-tune, and serve faster than ever.👉 Explore Token Factory🔗 Learn more

pg_lake integrates Iceberg and data lake files into Postgres, transforming Postgres into a stand-alone lakehouse system that supports transactions and fast queries on Iceberg tables. It can create and modify Iceberg tables directly from PostgreSQL, and directly work with raw data files in object stores like S3. pg_lake can be set up using Docker for an easy, ready-to-run test environment.

The CSS Custom Highlight API lets developers style arbitrary text ranges without modifying the DOM structure. Developers just have to define 'Range' objects that point to a specific character position in text nodes, group them by style type, and register them with the browser's highlight registry. The API is supported in all modern browsers. This post provides an example implementation of syntax highlighting using the CSS Highlights API.

Huge bursts of productivity concentrated in one industry are great because they create a huge number of new jobs and new opportunities, but this has an interesting spillover effect: the more jobs and opportunities created by the productivity boom, the more wages increase in other industries. These industries still have to compete in the same labor market. As the AI supercycle plays out, some goods and services will become orders of magnitude cheaper, while others will become more expensive - but we'll consume more of them anyway. This effect will even be seen within a single job - some work will see 10x throughput at 10x the quality, while other work, which must be done by humans, will command a wildly high wage.

AI has transformed how some developers work. We are just starting to see what the technology can achieve. We still haven't fully explored what current models can do. There are some concerning possibilities with the future of the technology, but for now, it is still in its 'wondrous' stage.

Adaptive Security—backed by OpenAI and a16z—stops AI powered social engineering through deepfake simulations, training, and risk scoring. Want to chat with a custom interactive deepfake of your boss? Book a demo

What matters in vulnerability reports is who submits them, what their incentives are, and whether they have actually verified their claims.

The Senate still needs to vote to confirm Isaacman as NASA administrator before the end of the calendar year, otherwise, he will have to resubmit conflict-of-interest paperwork.

Architectural debt on business and strategy layers can do more damage than architectural debt on technical layers.

EU tech regulations have fractured the once-unified, global Apple experience.

Younger developers will build on platforms that don't make them prove their worth through suffering.

Deepnote is a drop-in replacement for Jupyter that can scale to Deepnote cloud for real-time collaboration, Deepnote agent, and deployable data apps.