You know Wix Engineering on LinkedIn. Now come meet us on Instagram 👋 Follow Wix Engineering for the code, the people, the lessons learned, and everything engineering: instagram.com/wix_eng/

AI context engineering, 756K+ production messages, 6B+ daily media requests, zero-downtime database migrations, infrastructure migrations across 1,000+ MySQL servers, our upcoming meetup and more! Here’s what’s inside our latest @Wix Engineering newsletter - 🧠 From Autocomplete to Intelligence: How Language Became the Operating System of AI / @GiladBarkan AI can write code, use tools, and power autonomous agents - but underneath it all, it still predicts the next word. Explore how language became the operating system of AI, why Transformers and scale changed everything, and where today’s models still fall short. 🚦 Don’t Offer What Can’t Be Done: Deterministic Executability Gating for LLM Skill Selection at Scale / @AshkenaziOrtal, Vitalii Kloz & Mykhailo Ulianchenko Semantic matching alone isn’t enough in production. Across 756K+ messages, Wix’s executability gate filtered out 59.4% of matched skills, reduced skill context size by 90.5%, and saved 228.8M tokens by checking what users can actually do before skills reach the LLM. Now as a new paper on @arXiv. 🛠️ [Upcoming Meetup] Migrations at Scale: Code, Infrastructure & Automation / @Danaavital58 & Shmuel Mekonen Two migrations, two very different challenges: automating code changes across hundreds of codebases, and migrating 1,000+ MySQL servers across 160+ clusters in 30 days - with zero downtime. [Oct 28, at Wix Campus TLV] 🖼️ Architecting for 6 Billion Daily Requests: Inside Wix’s Media Platform / Tomer Salton What happens when image resizing needs to work across 300M+ websites and more than 6 billion media requests every day? Go behind the architecture powering real-time AVIF encoding, smarter caching, lower resize costs, and secure asset delivery at massive scale. 🎯 Attention is Expensive: Optimizing Context Windows for Precision and Cost / Maya Halevy More context doesn’t always mean better AI. Learn how attention limits, token waste, and the “Lost in the Middle” problem impact LLM accuracy - and how better context design can improve reliability while reducing cost. 🗄️ How We Built a Zero-Downtime Database Migration Service at Wix / Bar Shauli How do you move critical production databases between clusters without users noticing? Explore the Kafka, ProxySQL, and MySQL architecture behind Wix’s internal migration service - from streaming data to cutting over live traffic with zero downtime. Read it here 👇 wix.engineering/so/2dQ3DoCRl

[new @arxiv paper] There’s a huge gap between playing with AI skills in a demo and ⚙️ engineering them for production. In a simple setup, semantic routing feels like enough: the user asks something, the LLM matches the text to a skill description, and executes it. But in real-world production at scale (like @Wix Customer Care), semantic matching alone isn't enough. 🚩 Why? Because semantic matching only looks at what the user says, completely ignoring what the user is actually allowed or able to do. A user asking to "connect a domain" is a 100% semantic match for the Domain Skill. But if they don't have an active premium plan, executing that skill is a dead end. In a new arXiv paper by @AshkenaziOrtal, Vitalii Kloz, and Mykhailo Ulianchenko, we present Helpmate’s (Wix’s Customer Care AI agent) architecture: introducing a deterministic executability gate that evaluates account state before candidate skills reach the LLM. Tested across 756K+ user messages in production: 🎯 59.4% of semantically matched candidates were filtered out because the account state made them non-executable 📉 90.5% reduction in skill context size 228.8M tokens saved 🛡️ In a targeted replay of 1,000 risky conversations, removing the gate caused the model to select a non-executable skill in 7.8% of cases. Read the full paper: arxiv.org/html/2608.01050v1

Thank you @HebrewU_heb for highlighting Shmuel Mekonen's work, article and upcoming meetup: 1 - @HebrewU_heb - alumni.huji.ac.il/event/wix_… 2 - Article: 1,000 Servers. 160 Clusters. 30 Days. Zero Downtime - Migrating @Wix's MySQL Fleet to Graviton wix.engineering/post/1-000-s… 3 - Meetup: Migrations at Scale: Code, Infrastructure & Automation wix.engineering/migrations-a…

[Meetup] Large-scale migrations are never just about moving code or infrastructure. They're also about doing it without 🚩 slowing everyone else down. At @Wix, two very different migration challenges required two very different approaches: 1 - Rolling out automated code changes across hundreds of codebases, while keeping developers focused on day-to-day work. 2 - Moving 1,000+ MySQL servers across 160+ clusters, multiple regions, and two major platform transitions - all within 30 days and with zero downtime. Dana Avital and Shmuel Mekonen will share how these migrations were designed and executed: the architecture behind them, the automation that made them possible, the tradeoffs, the failures along the way, and the key decisions that helped them succeed. Join us on October 28 at 17:30 at Wix Campus TLV: wix.engineering/migrations-a…

[🧰 Workshop] Octocode for AI Agents & Developers Join our hands-on 🐙 Octocode workshop, led by @guy_bary, to learn how to build MCP servers, AI tools, and agents that can actually research and understand your codebase. You'll learn how to: 🛠️ Build MCP servers & skills 🔎 Research local code, GitHub and npm with Octocode 🧰 Build reliable, evidence-based AI tools 💻 Apply agentic development to your everyday workflow * When: October 12, 16:30-20:00 * Where: @Wix Campus, TLV Apply here: wix.engineering/workshop-oct…

If AI generates code faster than your team can review it, you haven't accelerated engineering. You've accelerated technical debt. At scale, the critical AI coding layer isn't generation. It's the infrastructure that gives agents context, guardrails and a reliable review loop:

AI can write code, use tools, and power autonomous agents. Yet at its core, it does one surprisingly simple thing: predict 🔮 the next word. So how did autocomplete become intelligence? In his new article, Gilad Barkan cuts through the weekly noise of new models and frameworks to explain why language became the operating system of AI. Learn how LLMs actually acquire intelligence, why Transformers and scale changed everything, and where today’s models still fall short. Along the way, he connects ideas from leading thinkers and AI pioneers, including @ylecun, @AndrewYNg, and @harari_yuval. No technical background required: wix.engineering/post/from-au…

Most AI failures 🚩 don't start with the model. They start with the context. We keep asking LLMs to reason over more prompts, more retrieved data, more instructions, more examples, and bigger context windows. But more context doesn’t automatically mean better answers. Sometimes it means higher costs, lower accuracy, and models that miss the most important part of the input. In her @Wix Engineering Conference 2026 talk, Maya Halevy breaks down what actually happens inside LLMs, including attention limits, token waste, and the "Lost in the Middle" problem. More importantly, she shows how to design context that actually helps the model perform better. Watch her session to learn how to: ✅ Improve model accuracy ✅ Reduce unnecessary token usage ✅ Structure prompts, instructions, examples, and retrieved data ✅ Build more reliable AI agents, RAG flows, copilots, and LLM-based products If you're building with LLMs, this is one of those talks that makes you rethink what "good prompting" really means: piped.video/fwI4p9l8ncs?si=PWNK…

The biggest shift in software engineering isn't a new programming language. It's a new way of thinking. 🧠 In this great talk, @tomenden explains why AI is pushing us beyond deterministic programming into a probabilistic world, what that means for modern engineers - and how to build reliable systems when your teammate is an LLM... Watch and read the full article: wix.engineering/post/the-end…

A migration can be running perfectly - and still be mathematically impossible to finish. If you migrate at 5,000 RPM while 15,000 new changes arrive every minute, you're not making progress. You're falling further behind. Before migrating a single record, do the math: can your pipeline actually outrun production? 👇

Happening tomorrow at @AWS Summit TLV! 🚀 @Wix's Ben Chen sharing practical lessons on building developer platforms that improve productivity - without sacrificing developer experience. A great session, a great speaker, and a great event by AWS. Good luck, Ben!

"Resize an image 🎨. How hard can it be?" Now do it across 300M+ websites and 6B+ media requests. Every day. Suddenly, every memory copy slows you down. Every cache miss repeats expensive work. Every resize burns CPU. And an encoder upgrade can break your tests - even when the image still looks perfect. In his new post, Tomer Salton reveals how @Wix Media Platform handles this scale: ✅ Making real-time AVIF encoding work across Go and C ✅ Cutting resize costs with intermediate “Master” images ✅ Turning different URLs with identical outputs into one cache hit ✅ Testing visual quality with SSIM and VMAF ✅ Serving previews and paid originals securely from the same asset Read the full story here: wix.engineering/post/archite…

Millions of users. Critical 🚩 production databases. Zero downtime. No room for mistakes. At @Wix, shared database clusters posed a serious reliability challenge: one problematic application could impact many others. The solution? 🚩 Move databases between clusters - without interrupting the applications using them. Bar Shauli shares how she and the Wix Data Infrastructure team built an internal migration service using Kafka and ProxySQL on top of MySQL to make these moves safely, transparently, and fully online at massive scale. From streaming data and handling partitioning challenges to cutting over live traffic - here's the engineering behind database migrations that users never notice: wix.engineering/post/how-we-…

[Meetup] What does it actually take to migrate code and infrastructure at @Wix scale? Two migrations. Two very different 🚩 challenges: 1 - Automating code changes across hundreds of codebases ➡️ without pulling dev teams out of their context. 2 - Migrating 1,000+ MySQL servers across 160+ clusters, multiple regions and two major platform changes ➡️ in 30 days, with zero downtime. @Danaavital58 and Shmuel Mekonen will go behind the scenes: the architecture, automation, tradeoffs, things that broke, and the decisions that made these migrations work. Join us at our upcoming meetup on October 28, 17:30, at Wix Campus TLV: wix.engineering/migrations-a…

Building APIs for AI agents? Design ❌ error responses to help them recover. When validation fails, return what went wrong, the expected schema, and a link to the specific docs needed to fix it. The agent can then use that feedback to repair its request. A practical example of Feedback-Driven Agents from @yoavabrahami. Read the full article:

What do you do when you lose an AirPod and can't access Apple's Find My app? If you're @denysdovhan, you build your own solution. Denys used Claude and the Bleak Python library to create a lightweight macOS utility that scans nearby Bluetooth devices and tracks their signal strength. About 20 minutes and 20 lines of code later, he just walked around his home with his MacBook and found the missing AirPod in a shorts pocket. A great example of how AI can help engineers turn a small everyday problem into a working solution in minutes. Read on @devua_official: dev.ua/en/news/inzhener-wix-…

AI coding tools don't fail at scale because the models aren't good enough. They fail because engineering organizations were built for humans - not agents. Claude Code, Cursor, Codex, and Copilot work great in a small repository. But across 2,500+ microservices and thousands of engineers, you need infrastructure that prevents broken PRs, lost context and AI-generated chaos. That's where @Wix's shift to AI-native engineering and xEngineers begins. In his article, Nitay Rabinovich explores: ⚡ Automated PR reviews 🤖 AI-powered build fixes 🧠 Agent-ready repositories 🔄 The evolution toward AI-native xEngineers If you're building software with AI, this is worth reading: wix.engineering/post/from-co…

5 billion records. 10 terabytes. Four weeks. 🚩 How do you migrate data at this scale without disrupting a live system - while millions of new changes keep arriving every day? In their new article, @avrahamr and Roy Noyman share the lessons behind one of @Wix’s largest data migrations. Read and learn how to: ✅ Uncover corrupted legacy data before it breaks the migration ✅ Detect parity gaps between V1 and V2 ✅ Prevent one failed record from blocking the entire pipeline ✅ Calculate the throughput needed to actually finish A practical playbook for anyone building or migrating large-scale production systems: wix.engineering/post/5-billi…

Puiki galimybė jauniems specialistams Vilniuje — ypač vertinga, kad programa suteikia ne tik žinių, bet ir galimybę dirbti su tikrais produktais bei komandomis.