The new OPFSWriteAheadVFS for wa-sqlite not only brings concurrent reads to SQLite on the web, but also makes write transactions faster!

We're at @stirtrek! Come say hi, grab a free tshirt, enter a giveaway for a Meta Quest 3S, and chat about building fullstack AI applications

PowerSync is proud to be sponsoring this year’s @stirtrek! And our own @devagrawal09 will be speaking on building Fast, Realtime, and Offline-First Apps with Sync Engines x.com/stirtrek/st...

How would you like a persistent browser database that: 💾 stores data durably in the browser 🪶 avoids Asyncify time and size overhead 🌐 skips COOP/COEP header requirements 🗂️ works across multiple tabs and workers 👀 supports simultaneous read transactions

If your AI app needs durable runs, resumable streams, background jobs, and shared history to be reliable, you're already partway to multiplayer. The next step is making that state collaborative so that your agents can align with not just a single user, but an entire team.

wa-sqlite now supports concurrent readers + a writer in Chromium, unlocking some massive performance gains for certain situations. How does this work?

Building AI apps isn't just about calling an API. It’s a fundamental shift in system architecture. We just published Part 1 of a new series on AI Engineering: The New Primitives. 🧠 LLMs as probabilistic functions 🛠️ Tool calling vs. APIs 🏗️ Context engineering

Most AI apps still package the experience as a private session: one user, one thread, one hidden context window. But real work is collaborative. If AI is going to participate in real workflows, its output needs to become shared state. Learn the why and the how in this post by @devagrawal09

Exciting news for SQLite on the web: wa-sqlite supports a new OPFSWriteAheadVFS virtual file system that unlocks much better performance by allowing multiple workers to read the database concurrently with a writer.

Talented Developers + Local First Data + AI = Absolute Gems🪎 The PowerSync AI Hackathon got some amazing submissions. Let's take a closer look at the bonus prize winners and see what a top tier AI app looks like

Most AI chat apps are built like demos. 📉 If you refresh the page or lose connection mid-generation, the response is gone. Why? Because we're treating LLM outputs as fragile streams instead of synced state. Learn how to fix your broken AI apps in this latest blog post by @devagrawal09.

What makes a good AI application? Let's learn from the top 3 projects of the PowerSync AI Hackathon

March was a big month for PowerSync. We shipped: 🗄️ SQL Server support beta 🟣 .NET SDK beta 📉 Query-driven sync for TanStack DB 🦀 Native Tauri SDK 💚 First-class Nuxt support Plus, our AI hackathon gave us a lot of useful signal from builders working on sync-powered AI apps.

🚀 @TauriApps developers, we fixed the storage problem. Introducing the PowerSync Tauri SDK (alpha): → Filesystem-based SQLite (data persists across app updates) → Native performance without WASM overhead → Automatic sharing of sync state across all windows

🌐 PowerSync backend support for SQL Server is now beta and production-ready for tested deployments. → CDC-based replication from existing SQL Server databases → Automatic schema change detection and handling → Support for Standard, Enterprise, Developer editions

💻 .NET developers, your sync SDK has arrived. The PowerSync .NET SDK is now beta with everything you need for offline-first applications: ✓ Stable APIs (breaking changes will be communicated clearly) ✓ Sync Streams for flexible sync strategies

Our final winner of the hackathon, securing the prize for "Best Submission using @Mastra" is....

The wins keep on coming! Presenting... the winner of the "Best Submission using @Neondatabase"

Meet the winners of the PowerSync AI Hackathon! 🏆 Huge thanks to everyone who participated in the PowerSync AI Hackathon — we were blown away by the submissions

🚀 Big update for @tanstack.com DB users! powersync-db-collection now supports Query-Driven Sync. No need to load 50,000 rows into memory when you only need to show 20. On-demand mode compiles your live TanStack DB queries into SQL for efficient data loading.

That’s a wrap on submissions for the PowerSync AI Hackathon. 54 projects are now in the hands of our judges. Thanks to every team and solo builder who joined, built fast, and pushed creative ideas forward. Winners soon.

How easy is it to go from a simple specification of a feature to a fully working implementation with AI? Turns out, pretty easy once you got PowerSync! youtu.be/9O-RUC9Rkl0...

Final day for the PowerSync AI Hackathon. If your project is working and the core idea is clear, submit it. A focused submission with a clear demo is enough. Submissions close tonight at 11:59 PM CT. Submit here: form.typeform.com/to/YR9U17Sn

PowerSync AI Hackathon builders, what are you working on? Reply with what you're building, who it's for, or a screenshot if you've got one. Would be great to see what people are shipping this weekend.

"How much of this can be local" should be the default question every dev asks. Most have gotten too comfortable with defaulting to remote hosting even the tiniest services. Glad to see a growing pushback for local-first. A nice mini-doc from @cultrepo.bsky.social www.youtube.com/watch?v=10d8...

If you're building this weekend for the PowerSync AI Hackathon, resist the urge to keep expanding the project. Get the core flow working. Make the value obvious. Make sure someone can understand the product quickly.

You can still start a PowerSync AI Hackathon project this week and submit something strong by Sunday. Keep the scope tight. Pick one user, one workflow, and one place where sync, offline support, or durable local state makes the product noticeably better.

With the deadline now on Sunday, this is a good time to simplify. The strongest submissions usually have: - a clear use case - a focused demo - an obvious reason PowerSync matters A smaller project that is easy to understand will go further than a bigger one that tries to do everything.

Update: we're extending the PowerSync AI Hackathon submission deadline to Sunday, March 22 at 11:59 PM CT. More time to finish strong, polish demos, and get submissions into good shape. If you're already building, keep going. If you haven't submitted yet, you've got the weekend.

A good PowerSync AI Hackathon project usually starts with a simple question: What AI product becomes much more useful when it keeps working across devices, handles weak connectivity well, or keeps state locally? That’s where the most interesting ideas tend to come from.

Looking for a PowerSync AI Hackathon idea? Ask your coding agent: "Give me 5 ideas for the PowerSync AI Hackathon. Focus on AI apps where offline support, cross-device sync, or durable local state clearly improves the product." Choose one and start building. Deadline: Friday, March 20.

🦀 $500 BONUS PRIZE for the Best Rust SDK Submission in the PowerSync AI Hackathon! Our pre-alpha Rust SDK is ready for builders who want to push the boundaries of local-first and AI. Build embedded systems, IoT apps, or edge-native agents — all powered by real-time sync.

Supabase gives you the Postgres backend and auth primitives that every app needs. PowerSync gives you the data sync layer that makes your app feel instant and survive offline. Together they're the stack for agentic apps that work everywhere—subway, airplane, rural connectivity.

Mastra gives you opinionated patterns for agent design, workflow orchestration, state management, and deployment that scale with your team. PowerSync slots in as the sync layer that makes that architecture work offline and across devices seamlessly.

Agentic apps iterate fast, and your database should keep up. Neon gives you serverless Postgres with instant branching — spin up a new DB per feature, per preview, per experiment.

Building agentic apps? Use TanStack Start for full-stack delivery, TanStack AI for type-safe tool calling, and TanStack DB for live incremental queries + optimistic mutations. Add PowerSync for offline-first sync, and your app keeps moving even when the network doesn’t.

Cloud inference bills add up fast. @cactuscompute changes this model — run inference on-device, skip the API costs entirely. Pair that with PowerSync's local-first sync and you've got agentic apps that are fast and cheap.

Got an idea for the next big AI app? Participate in the PowerSync AI Hackathon to compete for over $8k in prizes, including bonus prizes from our partners @supabase.com, @mastra, @tanstack.com, @neondatabase, and @cactuscompute. Submissions end on March 20th! Full details in reply

The PowerSync AI Hackathon starts today. Bring your favorite AI ideas to life and compete for over $8k+ in prizes, including bonus prizes from our partners @supabase @neondatabase @mastra @tan_stack @cactuscompute. Let the hacking being!

PowerSync now has first-class @nuxt_js support (alpha)! `@powersync/nuxt` adds auto-imported composables, built-in sync diagnostics, and optional Kysely so your Nuxt app can be offline-first + real-time.

PowerSync now has Agent Skills! Claude Code can now build real-time and offline-first apps for you. Give your favorite coding agents the ability to ship instant UX and reliable data sync without learning a new database language! $ npx skills add powersync-ja/agent-skills

The PowerSync CLI just went open-source and received an upgrade! - Support for self-hosted and Docker workflows - Browser based login flow - Validate Sync Stream definitions and migrate from Sync Rules - Plugin system - ...and much more!

Today we launched the beta of Sync Streams, our new partial sync architecture that is the successor to legacy Sync Rules. Sync Streams is a flexible format that lets you define exactly which data from your backend can sync to each client using simple SQL-like queries.

Want to make your app work while offline? Great news! All you need is your existing backend database and @powersync.com with some @tanstack.com DB for additional seasoning!

Finally took the time to set up local instances of Supabase and PowerSync. Surprisingly simple to get up and running actually. Also set up Cloudflare tunnels for them and the local Vapor server, so I can develop on real devices. #buildinpublic #indiedev #iosdev #swift

𝗔𝗻𝗻𝗼𝘂𝗻𝗰𝗶𝗻𝗴 𝘁𝗵𝗲 𝗣𝗼𝘄𝗲𝗿𝗦𝘆𝗻𝗰 𝗔𝗜 𝗛𝗮𝗰𝗸𝗮𝘁𝗵𝗼𝗻 🤖🧑‍💻 We are hosting a virtual hackathon where the challenge is to build innovative AI-powered software using PowerSync as a sync engine.

The new PowerSync Dashboard, powered by @tan_stack Router and Query, is now featured on the TanStack Showcase!

🚢 PowerSync Product Changelog — January 2026 Some highlights: ✅ SQL Server & Azure SQL support ✅ HIPAA-compliant PowerSync Cloud ✅ GRDB + SQLite encryption for Swift ✅ Better docs, faster Rust sync client as the default, and more Read the full update 👇

What if you could get all the benefits and capabilities of modern sync engines without giving up your existing tech stack? Learn more about how PowerSync makes building modern and agentic applications simple in our latest blog post “Unleashing the Power of Sync”

If your synced Swift app stores sensitive data on the device, you can now opt in to built-in SQLite database encryption in PowerSync’s Swift SDK.