If I don't respond to tweets or DMs it's because I don't know how to Twitter

The Phone Number Verification dev preview is launching with @deutschetelekom (Germany) and @Telkomsel (Indonesia), with more carriers coming. Check out the blog & fill out the interest form in the blog:

This is a differentiator for onboarding. The SDK is smart, too—it lets you check for carrier compatibility first. If the carrier isn't supported, you can fall back to your existing SMS flow. It's a perfect progressive enhancement.

The benefits are massive: ✅ Blazing fast: Logs users in seconds. ✅ Higher Conversion: No more "waiting for SMS". ✅ Lower Costs: Kills SMS AIT fraud. ✅ More Secure: No SMS to phish. ✅ Works on Wi-Fi: Unlike header enrichment, it's not dependent on a cellular data connection.

Firebase just dropped what we've been waiting for: Firebase Phone Number Verification (PNV), now in developer preview. This isn't just an autofill helper. It’s a new auth method that securely gets the phone number directly from the SIM via a carrier API. One tap. No code.

Developers, let's be honest: SMS OTP for sign-in is awful. It's slow, users drop off, it's vulnerable to phishing, and AIT (traffic pumping) costs are a nightmare. 😩 What if you could skip SMS entirely?

This is the future of building privacy forward AI. On-device processing for privacy, with a smart cloud fallback for compatibility. Read the full technical deep dive on how we built it here: https://t.co/SC4BoOi4tI #AI #WebDev #Privacy #GeminiAI #Firebase #OnDeviceAI #PWA

To make the AI's output useful, we enforced a strict JSON schema. This gives us clean, predictable data (descriptions, categories, colors) to work with in the app—no messy string parsing needed!

But what if a browser doesn't support an on-device model? We implemented a hybrid AI approach. The Firebase AI SDK automatically falls back to a cloud model (Gemini 2.5 Flash-lite), but ONLY after transparently asking the user for permission. Control stays with the user.

This on-device-first approach means: ✅ Your photos never leave your machine. ✅ The app works completely offline. ✅ User trust is built into the foundation. We built it as a Progressive Web App (PWA) to reinforce this offline-first, privacy-centric design.

The problem: Most AI features send your data to a server for processing. For an app that handles personal photos, that's a huge privacy concern. The solution: Bring the AI to the photos, not the other way around. We used Gemini Nano to run analysis directly in the browser.

Building AI features is cool, but what about user privacy? 🤔 We built a smart photo organizer that runs AI ON your device, not in the cloud. Your personal photos stay personal. Here’s how we did it with a privacy-first approach. 🧵

Practical AI examples to level up your app! ✨ Firebase AI Logic and Gemini can auto-suggest star ratings while a user types. Efficient, secure, & magical. #firebase #ai

Join us TODAY on the Firebase After Hours livestream as @NoheDev and I chat with @riggaroo from the Android DevRel team how they built Androidify using Firebase AI Logic! https://t.co/ELODlk50vz

Live coding alert! 🚨 Join us this week on #FirebaseAfterHours as @peterfriese and @NoheDev team up with @rodydavis from Firebase Studio for some serious vibe coding. 🎧💻 See Firebase Studio in action and get your questions answered live → https://t.co/7ob2h7u8Y6

👀 See Imagen 3 in action → https://t.co/kabiO0IGdC In this month's Deep Dive, @NoheDev demos image generation via Vertex AI in Firebase and presents a coding challenge with Sparky! 🔥

Obtaining verified phone numbers for your app on Android has never been easier. With Firebase you'll soon be able to get a user's verified phone number on Android with a single call. ⬆️ Usability: Users no longer have to enter their phone number - it's shared with your app

⚡️ Level up your app's personalization! ⚡️ Learn how to use Firebase Remote Config's custom signals to deliver tailored experiences based on real-time user data like battery level! 🔋📱 Dive into my tutorial: https://t.co/UfEUxZE7RG #Firebase #RemoteConfig #MobileDev

We ask this all our guest - “What is your preferred code indentation style” - most people indent 2 or 4 spaces (or use tabs), but my co-host @NoheDev really knocked it out of the park with “Fibonacci indentation”: https://t.co/9KWJJacif5

Recipe generator upgrade! 🥳 Imagen3 + Genkit = 💥 Visuals & UX improved! 🤩 Love Genkit’s flexibility for model switching. ✨ Time saver! ⏳ #GenerativeAI #Genkit #Firebase

We're headed live to answer all of your burning demo day questions: https://t.co/uuH66BWUpZ Come join us in the chat!