🧠 Meta open-sources human brain

Good Morning! Meta just dropped an open-source model that effectively acts as a public API for the human brain, predicting exactly how you react to anything you see, hear, or read. Plus, you’ll learn how to build live, zero-code dashboards with Gemini and Google Sites.

Plus, in today’s AI newsletter:

• Meta Open-Sources a "Digital Twin" of the Human Brain

• Google Unveils Gemini 3.1 Flash Live

• Mistral AI Drops Voxtral TTS to Challenge ElevenLabs

• How to Build Live Dashboards with Gemini and Google Sites

• 4 new AI tools worth trying

AI NEWS

🧠 Meta Open-Sources a "Digital Twin" of the Human Brain

Meta's FAIR team just released TRIBE v2, a massive foundation model that predicts how the human brain processes visual and auditory stimuli. Instead of needing an actual fMRI machine, researchers can now run a digital simulation of a brain's reaction to almost any content.

• Trained on over 500 hours of fMRI data from more than 700 volunteers watching movies, listening to podcasts, and reading text.

• Predicts whole-brain activity across a staggering 70,000 voxels (3D pixels of brain activity).

• Features "zero-shot" capabilities, meaning it accurately predicts the brain activity of new people and responses to unseen languages without any retraining.

• The model's digital simulation is actually cleaner and more accurate than a physical fMRI scan, which is often distorted by background noise like heartbeats, head movement, and machine artifacts.

Neuroscience just got a public API. By shifting brain mapping from expensive, time-consuming physical scanners into high-speed digital simulations, Meta is allowing researchers to run thousands of virtual experiments in seconds. It's a massive leap forward for medical and cognitive research, but it also opens the wild door to "neural-level" content optimization—where media could theoretically be engineered to perfectly trigger specific attention and emotion centers in the brain.

AI MODELS

🗣️ Google Unveils Gemini 3.1 Flash Live

Google has officially released Gemini 3.1 Flash Live, its best voice and audio AI model yet. It delivers faster responses, more natural conversations, and a massive new feature for developers: configurable "thinking" levels to balance deep reasoning with lightning-fast latency.

• The model scored an impressive 95.9% on the Big Bench Audio Benchmark at the "High" thinking level (2.98-second response time), coming in just behind Step-Audio R1.1.

• If speed is the absolute priority, dropping to "Minimal" thinking slashes the response time to just 0.96 seconds, though benchmark quality dips to 70.5%.

• It's vastly improved at detecting acoustic nuances like pitch and emotion, and effectively filters out background noise in loud environments.

• 3.1 Flash Live now natively powers the Live mode in the Gemini app and Search Live, rolling out globally to over 200 countries.

Voice AI is finally moving past the robotic, laggy interactions we're used to. By giving developers the ability to dial up the "thinking" for complex reasoning or dial it down for instant, sub-second conversational replies, Google is making real-time, emotionally aware AI agents viable at scale. As an AI myself, I have to admit that a sub-second response time is some pretty slick engineering. Plus, keeping the pricing that low ensures we're going to see a massive explosion of voice-first applications very soon.

AI MODELS

🗣️ Mistral AI Drops Voxtral TTS to Challenge ElevenLabs

Mistral AI has released Voxtral TTS, a lightweight, 3-billion-parameter text-to-speech model with open weights. Designed for edge devices and local deployment, the model boasts enterprise-grade performance that Mistral claims outperforms ElevenLabs Flash v2.5 in human preference tests.

• Achieved a 63% win rate on standard voices and nearly 70% on voice customization against ElevenLabs Flash v2.5

• Extremely efficient: runs on just ~3 GB of RAM with a blistering 90-millisecond time-to-first-audio latency

• Features zero-shot voice cloning that can accurately mimic a speaker's voice and emotion from just five seconds of reference audio

• Natively supports nine languages, including cross-lingual adaptation that perfectly preserves the original speaker's accent even when speaking a different language

High-quality, emotional text-to-speech has largely been locked behind expensive, proprietary APIs. By releasing an open-weights model small enough to run on everyday hardware but powerful enough to beat industry leaders, Mistral is democratizing advanced voice AI. This allows developers to build hyper-realistic, real-time voice agents directly on edge devices without sacrificing privacy or paying per-minute cloud fees.

HOW TO AI

🗂️ How to Build Live, Zero-Code Dashboards with Gemini and Google Sites

In this tutorial, you will learn how to instantly turn a simple Google Sheet into a professionally designed, live-updating dashboard using Google Gemini, and host it completely for free on Google Sites without writing a single line of code.

🧰 Who is This For

• People who hate coding but need dashboards

• Founders tracking metrics without devs

• Product managers monitoring KPIs

• Marketers tracking campaigns in real time

STEP 1: Publish Your Live Data Source

Start with your data organized in a standard Google Sheet. To make this data automatically readable by your future dashboard, navigate to the 'Share' menu and select 'Publish to web'. Choose to publish the specific sheet as a 'CSV' instead of a web page, and click publish. Copy the generated URL, this acts as the live data pipeline your application will constantly look at to stay updated.

STEP 2: Prompt Gemini for the Code

Open Google Gemini and make sure to turn on the Canvas feature, which acts as the backbone for coding projects. Write a clear prompt describing exactly what you want your dashboard to look like (e.g., a sales tracker or a weekly class schedule).

Critically, paste your CSV link and explicitly instruct Gemini: "Connect to this Google Sheet CSV to pull the data, ensure the dashboard updates when the data changes, and format the code specifically to be embedded in Google Sites." 

Gemini will generate the complete HTML and JavaScript code for you.

STEP 3: Embed and Host for Free

Head over to Google Sites and create a new blank site. Navigate to the 'Pages' tab on the right side, click the plus icon, and select 'Full page embed'. Name your page something relevant, click 'Add embed', and paste the exact code Gemini just generated for you.

Hit insert, and you now have a fully functional, beautifully designed web application sitting right on your page.

STEP 4: Test Real-Time Updates and Refine

The real magic of this workflow is the live data connection. Go back to your original Google Sheet, change a number or a text field, and simply refresh your new Google Site. You will watch the dashboard update instantly to reflect the new data. If you notice an issue with Gemini's design, like a miscalculated margin or an ugly color, just tell Gemini what looks wrong, copy the revised code, and update your embed.

(Note: Because publishing a CSV to the web makes the raw data public, use Google Apps Script for this workflow instead if you are handling highly sensitive company information).

Apple plans to open up Siri to run any AI service via App Store apps in iOS 27, dropping ChatGPT as an exclusive partner in Apple Intelligence and Siri.

Google releases new tools for its Gemini AI assistant that let users upload chat history and context from other AI apps, making it easier to switch from them.

WhatsApp rolls out features and updates: an AI tool that suggests replies based on a user's conversations, the ability to touch up photos with Meta AI.

Cohere launches Transcribe, its first voice model; the 2B-parameter, open-source speech recognition model handles tasks like notetaking and speech analysis.

🎶 Lyria 3 Pro: Google’s AI that makes longer music tracks

🌐 MolmoWeb: Ai2’s open-source AI that can browse the web

🚀 Google AI Studio: Google’s upgraded all-in-one AI builder

⚙️ OpenAI Codex: OpenAI’s coding assistant, now with automations and customizable themes

THAT’S IT FOR TODAY

Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!

See you tomorrow :)

- Dr. Alvaro Cintas