😸 Real-time translation is finally real

Welcome, humans.

Is anyone else getting confused by the Alphabet Soup of AI Gadgets and Gizmos we’re all expected to keep track of? AI Skills, Projects, Gems, Custom GPTs, Agents, Plugins, and now “Loops”? How are us normies expected to keep it all straight without going loop-de-loop ourselves?

Well, turns out all of those solve slightly different problems. And the hard part is knowing which one belongs in your workflow before you spend 45 minutes building a custom assistant for a task that basically just needed a digital sticky note.

Later today @ 10AM PT | 1pm ET on Neuron LIVE, we’ll break down the best practices for when to apply all of those different whozits and whatzits: when to use a project, when to use a skill, even when to run a loop to let an agent loose to go do something all on its own.

Just click the image below to open the tab ahead of time, then click “Notify Me” to get notified right when we start. Or go to LinkedIn and save the tab ‘til then. TTYIALB (“talk to you in a little bit”; yeah i just coined an initialism, come join the live to yell at me).

Click the image to go to YouTube, then on YouTube, click “Notify Me” to get pinged right when we start.

Here’s what happened in AI today:

😸 Google launched real-time voice translation in 70+ languages
📰 OpenAI is reportedly planning major price cuts to compete with Anthropic ahead of both IPOs.
📰 JPMorgan is deploying AI agents that run autonomously for hours, boosting private banking sales 20%.
📰 Mustafa Suleyman called Anthropic's Claude consciousness approach "really, really dangerous."

…and a whole lot more that you can read about here.

Hey: Want to reach 700,000+ AI-hungry readers? Advertise with us!

P.S: Love robots? We’re starting a new robotics newsletter! Sign up early here.

😸Google Just Made Real-Time Translation Actually Work
You figured out AI. Now you're everyone's IT desk.
🎓 AI Skill of the Day: Interview Anyone, in Any Language, Right Now
🍪 Treats to Try
NEW from The Neuron: Microsoft AI’s Plan for “Humanistic Superintelligence” Post OpenAI
📰 Around the Horn
Secure Your Code, Pipelines, and AI Agents at Scale
🧩 Thursday Trivia
A Cat’s Commentary

😸Google Just Made Real-Time Translation Actually Work

You've been living the translator nightmare. Google just ended it.

You're presenting to an international client. Half the room speaks Japanese, the other half German. Your interpreter is a $500/hour human trying to keep up while you're mid-flow. There's lag, there's lost nuance, there's someone staring at their phone.

That's the world Google just made slightly embarrassing.

Gemini 3.5 Live Translate launched Monday, bringing near real-time, natural-sounding speech translation across 70+ languages. No choppy pauses. No robotic monotone. It actually sounds like a person.

Here's what happened:

Google opened the Gemini 3.5 Live Translate model to developers via the Gemini API and AI Studio.
The model handles real-time speech-to-speech translation as a continuous stream, not chunk-by-chunk.
It auto-detects language switches mid-sentence without any manual configuration.
It's already live in Google Translate (iOS and Android, with headphones) and rolling out in private preview for Google Meet.
Use case demo: one speaker presents to a global room, attendees scan a QR code and listen in their preferred language simultaneously.

Why this matters:

This is a bigger deal than it sounds. Real-time translation has been technically possible for a while, but Google's version is the first that sounds conversational rather than mechanical. The model handles cross-language streaming, meaning it doesn't wait for you to finish a sentence before it starts translating. That closes the one gap that's kept live translation from being usable in real meetings.

For knowledge workers with any international exposure, the productivity unlock here is real. Think global all-hands, client pitches, or vendor negotiations where you've historically needed an interpreter or just... hoped for the best.

Our take:

The API access is the sleeper story here. Developers can now bake real-time translation into anything: customer support tools, conferencing software, live event apps. The consumer demo is cool, but the infrastructure play is what changes industries. One open question: how well does it handle domain-specific vocabulary, legal or technical terms, for example, where a mistranslation carries real stakes?

And btw, did anyone see SemiAnalysis' thread today? They ran long coding tasks until they hit the weekly limits on both the $200/month Claude Max and OpenAI plans and found subscribers are getting way more than the assumed ~$2,000/month in token value. So before OpenAI slashes prices...you might already be getting a pretty good deal.

FROM OUR PARTNERS

You figured out AI. Now you're everyone's IT desk.

Come to Sauna instead 🧖

You figured out MCPs, APIs and the prompts that actually work. But every time a teammate wants in, you're back to square one - re-authing the same tools resending the same skills.

Sauna's new Spaces puts an end to that.

Invite your team and they inherit everything you've built - tools authed, integrations connected, skills from every session you've run. The moment they open it, they can see how you built each campaign, dashboard, scheduled task.

Try Sauna – Your Second Brain, that becomes a hive mind when you’re ready

🎓 AI Skill of the Day: Interview Anyone, in Any Language, Right Now

Most companies either outsource foreign-market user research or skip it entirely. Language is the wall. Live Translate just knocked it down.

Set up an interview with a non-English-speaking customer. Run Gemini 3.5 Live Translate in the background via Google AI Studio or Google Translate on your phone. You speak, they speak, and you take notes in real time as if they're talking your language. No translator in the loop. No "we'll follow up once we get it transcribed."

You're getting raw signal from a customer segment most teams have never been able to reach directly.

If you want to take it further, the Gemini Live API exposes the same model as a real-time audio stream — input goes in, translated audio comes out, continuously. The build pattern: capture mic audio, send chunks to the API with a source/target language config, pipe the translated stream to your output layer. It slots into any existing voice feature architecture. Google has working examples in AI Studio worth pulling apart first.

End result: you can add live translation directly into a customer support tool, a user research platform, or a live event app — without touching a third-party translation service.

Total AI beginner? Start here (goes with this video).

Have a specific skill you want to learn? Request it here.

🍪 Treats to Try

*Asterisk = from our partners (only the first one!). Advertise to 700K+ readers here!

*For the person everyone asks about AI: Dell Pro Max with GB10 brings NVIDIA Grace Blackwell power, 128GB memory, and DGX OS 7. Explore Dell Pro Max with GB10.
Concentrate AI gives your team one API to route work across 130+ models, track token spend, and keep apps running when providers fail (raised $5.1M); pay as you go, 0% service fee on token usage.
Tamadoggo keeps a living journal for your pet — logs walks, vet visits, and health patterns, then writes you a monthly letter about your pet's year so far —free to start.
Veltrix connects to QuickBooks, Xero, Shopify, or Square and lets you ask plain-English questions about your business finances, getting back real answers with a suggested next step —free trial.
SellerAI runs your Shopify or eBay store on autopilot, handling sourcing, listings, pricing, ads, and fulfillment with minimal input from you —free to start.
AgentOS by SapienX gives you a single control dashboard to manage, monitor, and approve actions across multiple running agents at once —free, open-source.
OrchestraML takes your machine learning goal (e.g. "predict churn") and runs the full pipeline — finding data, cleaning it, building and evaluating a model — with human approval checkpoints at every critical step —free to start.
Agentcad lets your coding agent design and export real 3D CAD files (STEP, STL, GLB) by describing what you want to build in plain text —free, open-source.

NEW from The Neuron: Microsoft AI’s Plan for “Humanistic Superintelligence” Post OpenAI

Microsoft dropped seven new AI models at Build, then Mustafa Suleyman sat down with Corey to explain the larger plan: making Microsoft self-sufficient in AI (meaning moving beyond relying on OpenAI) in order to build “humanistic superintelligence.” What is that? Gotta watch to find out!

Special thanks to Beyond Trust for sponsoring this video!

Pick your fave platform to watch/Listen: YouTube | Spotify | Apple Podcasts.

Also new from The Neuron:

GPT memory, GPT-5.6 rumors, Hermes, Codex plugins, and MAI-25 — our latest roundup of the biggest AI product rumors, model updates, and agent launches.
Claude Code creators explain agent loops — Boris Cherny and Cat Wu walk through how to keep coding agents moving without losing control.
Everything (AI) Apple announced at WWDC 2026 — Siri AI, Apple Foundation Models, Private Cloud Compute, Visual Intelligence, AI apps, and more.
I Think Apple Just Won AI: Here’s Why — the developer-side case for why App Intents, Foundation Models, Core AI, and Xcode agents could finally make the iPhone the ultimate AI platform… and how to use them.
Managed Agents Explained — a basic guide to how managed agents work, why they matter, and where they fit in the next wave of automation.📰 Around the Horn

📰 Around the Horn

This feels about right; super fascinating that GPT 5.5 (their latest) beat Claude Fable 5 (the “Mythos” class model from Anthropic)

UC Berkeley RDI’s Agents’ Last Exam showed frontier agents still struggling on real professional workflows, with OpenClaw scores making the harness look almost as important as the model.
OpenAI is considering drastic price cuts on tokens (the unit AI companies use to bill you) as it braces for a pricing war with Anthropic, per WSJ and both companies are heading toward IPOs and neither wants to blink first.
JPMorgan is rolling out AI agents that run autonomously for hours at a time, saying the tech has already driven a 20% bump in private banking sales and could eventually let each banker cover 50% more clients.
Microsoft blocked employees from accessing Claude Fable 5 through internal GitHub Copilot after Anthropic updated its data retention rules, with legal teams reportedly reviewing whether flagged prompts could expose customer data.
SpaceX said building orbital AI data centers won't require new technology, claiming the hardware already exists in Starlink V3 satellites, ahead of the company's expected $1.75T IPO this week.
ASML became Europe's most valuable company ever, hitting a $674B market cap as analysts bet that demand for its chip-making machines (which are required to produce the most advanced AI chips) will keep climbing.