Daily Tech News - 2026-02-11

The era of "vibecoding" is rapidly maturing into a landscape of massive agentic models and hardened infrastructure as the industry grapples with the fleeting nature of the first wave of AI-generated apps.

Leading the technical charge is Z.AI’s GLM-5, a staggering 754-billion-parameter MIT-licensed model that doubles the size of its predecessor. Optimized for "agentic engineering," GLM-5 is designed for autonomous tool use and complex long-range planning. This shift toward functional autonomy is mirrored in the OpenAI API, which now supports "Skills"—essentially zippable capabilities that can be sent as inline data to give models immediate, specialized powers. However, this technical leap comes with a reality check: while AI enables anyone to build, many "vibecoded" apps built by non-programmers last year have already vanished, proving that creation is only half the battle in a world where context windows are a dilutive resource rather than simple RAM.

In the professional ecosystem, Vercel is doubling down on enterprise-grade control. The platform recently moved Vercel Flags into public beta and introduced advanced egress filtering for its Sandbox, allowing for granular SNI and CIDR-based network policies. Interestingly, Vercel is also using agents to solve a human problem: their "Community Guardian" system attempts to scale community management while respecting the fact that human communities are "non-fungible" and cannot be manufactured or replaced like discrete units.

The day’s news also highlights the enduring nuances of the craft. Whether it is Zig moving to remove ambient IO to enforce explicit engineering, the mathematical precision required to solve the orthogonal Procrustes problem, or the nostalgic milestone of Sony’s final CRT (released twenty years ago this week), the message is clear: even in an age of agents, the details of performance, safety, and history still define the cutting edge.

Featured Articles

advanced-egress-firewall-filtering-for-vercel-sandbox-5ff85d8c

Vercel Sandbox can enforce egress network policies through Server Name Indication (SNI) filtering and CIDR blocks. Outbound TLS connections are matched against your policy at the handshake, unauthorized destinations are rejected before any data is transmitted.

Keywords: egress firewall, vercel sandbox, firewall policies, based firewall, sandbox vercel, peeking firewall, unrestricted internet, outbound tls, firewall rules, firewall
Source: vercel.com

aligning-one-matrix-with-another-76b50962

The orthogonal Procrustes problem is to find a rotation matrix that lines up B with A. The Frobenius norm is the 2-norm if we straighten the matrices into vectors of dimension n². The solution involves singular value decomposition (SVD)

Keywords: orthogonal procrustes, orthogonal matrices, solving orthogonal, rotation matrix, orthogonal matrix, constraint orthogonal, matrix minimizing, column orthogonal, solution orthogonal, orthogonal
Source: johndcook.com

communities-are-not-fungible-a8ff7940

There's a default assumption baked into how Silicon Valley builds products. It tracks against how urban planners redesign neighbourhoods. Communities are not interchangeable, and if you "lose" one, you can manufacture a replacement. The value of a community isn't in the people as discrete units.

Keywords: communities fungible, treating communities, existing communities, value fungible, communities interchangeable, communities, community survive, communities don, community structure, communities communities
Source: joanwestenberg.com

glm-5-from-vibe-coding-to-agentic-engineering-0d691d5e

GLM-5 is a huge new MIT-licensed model: 754B parameters and 1.51TB on Hugging Face. It's twice the size of GLM-4.7 which was 368B and 717GB (4.5 and 4.6 were around that size too)

Keywords: agentic engineering, coding agentic, glm vibe, agentic, seen agentic, agents, glm 368b, size glm, pelican riding, agents demo
Source: simonwillison.net

glm-5-is-live-on-ai-gateway-962aaecf

GLM-5 from Z.AI is now available on AI Gateway. It adds multiple thinking modes, improved long-range planning and memory. It's particularly strong at agentic coding, autonomous tool use and extracting structured data.

Keywords: zai glm, ai gateway, glm ai, model zai, zai, access glm, gateway provider, intelligent provider, gateway model, gateway provides
Source: vercel.com

how-do-i-suppress-the-hover-effects-when-i-put-a-win32-common-controls-listview-in-single-click-mode-75355264

A customer had a Win32 common controls ListView in single-click mode. This has a side effect of enabling hover effects: When the mouse hovers over an item, the cursor changes to a hand, and the item gets highlighted in the hot-track color. How can they suppress these hover effects while still having single- click activation?

Keywords: suppress hover, effects mouse, enabling hover, mouse hovers, notification suppress, click activation, single click, hover effects, hovers item, suppress hot
Source: devblogs.microsoft.com

keeping-community-human-while-scaling-with-agents-31991599

Vercel built a two-tiered agentic system, internally referred to as our Community Guardian. The first tier handles autonomous operations, running independently without constant oversight. The second is c0, our research layer.

Keywords: agentic workflows, tiered agentic, scaling team, community guardian, scaling agents, guardian operations, community management, workload team, guardian manages, guardian tier
Source: vercel.com

last-year-all-my-non-programmer-friends-built-apps-a0cc03c7

Last year, all my non-programmer friends were building apps. Yet today, those apps are nowhere to be found. The app-building AI services still exist, but their customers have paused their subscriptions.

Keywords: building apps, app building, apps, apps today, developing app, programmer friends, app ai, app phone, today apps, apps followed
Source: idiallo.com

link-checking-with-generative-ai-cddf19d2

The vibecoded agentically engineered webapp is very simple. It checks the Web site to see if all its links work, and it sends back a report on any broken links. The user can review the history of these checks or trigger a new check.

Keywords: breaking links, broken links, broke links, link checking, broken link, link checker, cloudfront broke, removing cloudfront, needing cloudfront, links home
Source: natemeyvis.com

matrix-ain-t-it-chief-a35f689e

Please wait a moment while we ensure the security of your connection. Making sure you're not a bot! Taking a moment to check your connection is secure. Back to Mail Online home. Back To the page you came from.

Keywords: bot loading, connection images, webp images, loading, images dec93748b65b, images 9cc5192d5d8b, loading wait, webp, bot, images images
Source: xeiaso.net

Many years ago I modified my shell environment on our servers so that it wouldn't report the currently logged in users, show the motd, or tell me my last login details. But I kept the 'uptime' line: $ ssh cs.toronto.edu 22:26:05 up 209 days, 5:26, 167 users, load average: 0.47, 0.51,0.60 : apps0.cs ;. I've made it so that I skip running 'upt time' if I'm logging in from home and the load on the machine I'm logged in to is sufficiently low to be uninteresting.

Keywords: logins report, login report, ssh logins, logging unix, shell profile, logged users, treat login, users motd, lot ssh, login sessions
Source: utcc.utoronto.ca

on-screwing-up-fabe382a

The most important thing is to control your emotions. If you’re anything like me, your strongest emotional reactions at work will be reserved for the times you've screwed up. There are usually two countervailing emotions at play: the desire to defend yourself, find excuses, and minimize the consequences.

Keywords: lie colleague, messed testing, conceal mistake, ashamed mistake, confess guilt, screwed colleague, ashamed handled, handle mistakes, mistakes concealment, screwing shameful
Source: seangoedecke.com

package-management-consulting-f85e4486

If you’re building or running a package manager, registry, or dependency tooling, I can probably help. I’ve documented the design patterns, tradeoffs, and failure modes across dozens of ecosystems. Ecosyste.ms tracks 13+ million packages across 50+ registries, 290 million repositories, and 24 billion dependencies.

Keywords: package managers, package management, package manager, dependency landscape, ecosystem package, dependency practices, dependency tooling, package registry, dependency strategy, packages
Source: nesbitt.io

programming-aphorisms-187a6338

This is a personal, descriptive post, not a prescriptive post for you. The background here is that Zig is in the process of removing ambient IO capabilities. In the next Zig version, you’ll have to thread grotesquestd.process.EnvironMap.

Keywords: meta programming, process environ, zig compiler, zig process, program environment, from_environment convenience, accumulated deconstruction, zig adapted, getenvvarowned zig, process coding
Source: matklad.github.io

proving-what-s-possible-3cd353ac

When "what CAN happen" is as important as "what SHOULD happen" Proving What's Possible is a way to show that a system is safe and liveness can be combined. It's possible to lose 100 dollars betting on slot machines, even if you only bet one dollar at a time.

Keywords: temporal logic, temporal operators, state reachability, formal methods, generally temporal, valid transitions, database state, temporal, linear temporal, properties systems
Source: buttondown.com

quoting-andrew-deck-for-niemen-lab-975b63ab

Built in-house and known internally as the “Manosphere Report,’ the tool uses large language models (LLMs) to transcribe and summarize new episodes of dozens of podcasts. An AI-generated report was an essential tool in the Times’ coverage.

Keywords: manosphere report, podcasts manosphere, report tool, inboxes journalists, ai tool, manosphere recent, ai generated, track manosphere, manosphere, journalists
Source: simonwillison.net

skills-in-openai-api-391fc5c4

You can now use Skills directly in the OpenAI API with their shell tool. You can zip skills up and upload them first, but I think an even neater interface is the ability to send skills with the JSON request as inline base64-encoded zip data.

Keywords: skills openai, openai api, skills_in_api, api openai, examples skills_in_api, script openai, skills_in_api md, skills json, use openai, openai
Source: simonwillison.net

the-last-sony-crt-ever-made-36f5bfa5

On February 14, 2006, Sony announced its final CRT television, the KD-34XBR970. The last Sony CRT TV was a 34-inch model, with high definition and a 16:9 screen. At the same time, Sony also released three conventional 4:3 standard definition models.

Keywords: sony crts, sony crt, crt tv, crt television, 34xbr970 sony, tv 34, crt technology, model sonys, sony announced, crt based
Source: dfarq.homeip.net

thinking-about-context-eba1b0de

The context window is one of the new fundamental ideas in programming. It's tempting to think of context by analogy to RAM, but this is dangerous. Adding to the context has a quadratically dilutive effect on it.

Keywords: choosing context, context context, context analogy, using context, context windows, resource context, context useful, context window, remember context, context free
Source: natemeyvis.com

unresponsive-buttons-on-my-fastest-hardware-ever-52d8759b

clicking on buttons results in slight but perceptible delays. When clicking on that button, even on a fast connection, my brain glitches for a second. I want (perceptibly) instant feedback. If something is happening, tell me!

Keywords: button fast, onclick async, unresponsive buttons, delay clicking, delay responsiveness, imperceptible delay, buttons fastest, slight delay, button spa, clicking wait
Source: blog.jim-nielsen.com

vercel-flags-is-now-in-public-beta-40f7ffdf

Vercel Flags is a feature flag provider built into the Vercel platform. It lets you create and manage feature flags with targeting rules, user segments, and environment controls. It is priced at $30 per 1 million flag requests ($ 0.00003 per event)

Keywords: vercel flags, verceladapter flags, vercel dashboard, flags feature, dashboard flags, feature vercel, feature flags, flags js, shownewfeature flags, backends vercel
Source: vercel.com