The Big Story

Google just launched Gemini 3.7 Flash, a new AI model designed especially for coding and building AI agents. The announcement calls it the company’s most intelligent workhorse model yet for coding and agents. Think of an AI agent like a helpful digital assistant that can carry out several steps in a row without you having to remind it every time. Instead of just answering one question, it can plan, use tools, and finish longer jobs across apps like Gmail or Google Docs. The model costs half as much as the previous version through the end of the year, which means more people can experiment without worrying about big bills. It is available at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens until December 31.

This change matters for anyone who likes building things or wants help with school projects, creative work, or daily tasks. A student could ask it to organize notes, draft an essay outline, and then check a calendar for deadlines—all in one go. The model thinks more diligently, putting extra effort into multi-step planning and tool calls so it produces a more disciplined execution with fewer retries. For future careers, learning to work with these agents now gives you an edge in fields like design, writing, or even game development.

You can try it today if you have Google AI Pro or Ultra through the Gemini app on your phone. Open the app, look for Gemini Spark, and give it a multi-step request like “Plan a weekend study schedule and add it to my calendar.” Even the free Gemini app lets you test similar features and see how the model thinks through problems step by step. Google also made the model available in the Gemini Enterprise Agent Platform, the Gemini Enterprise app, the Gemini API via Google AI Studio, and Android Studio. Source: x.com

Explain Like I'm 14

You know how when you’re texting, your phone sometimes guesses the next word before you finish typing? Now picture that same idea, but the guesses get much bigger. Instead of one word, the AI guesses what the next several actions should be. It looks at your request, decides which tools or apps it needs, and lines up the steps like a to-do list. Then it actually carries those steps out, checking its work as it goes.

The model puts extra effort into planning and tool use so it makes fewer mistakes and needs less fixing from you. That’s why it feels like it can handle longer, more complicated jobs without constant hand-holding. In practice this shows up when the AI works across Google Workspace apps such as Gmail, Google Calendar, and Google Docs in a single conversation. It can break a big request into smaller parts, call the right tools for each part, and keep track of what it has already done. The result is an assistant that stays on track even when the task stretches over many minutes instead of a single reply.

Cool Stuff & Try This

Talk to your music studio like it’s a bandmate Suno Studio 2.0 added a chat feature that lets you describe instruments or sounds in plain words and have the AI create them for you. You can also import MIDI files and export high-quality audio without limits on the Premier plan. This is perfect if you make beats, write songs, or just want to experiment with music on your laptop. Go to suno.com, start a new project in Studio 2.0, and type something like “create a soft piano with reverb that sounds like rain” to see what it builds. The chat turns the digital audio workstation into something closer to a bandmate that understands plain-language requests. Premier subscribers get the full set of features including unlimited 32-bit exports and MIDI import. Source: the-decoder.com

ChatGPT now remembers what you’ve been doing on your computer The new Computer History feature in the ChatGPT desktop app lets the AI recall your recent activity across apps and websites so conversations feel more natural. You don’t have to re-explain everything every time. If you’re working on a project, it can pick up right where you left off. Download the desktop app from openai.com, enable the feature in settings, and try asking it to continue something you were just doing in another window. The update makes future interactions more personalized because the model keeps a running record of what you opened and edited. Source: x.com

Quick Bits

Post and accept gigs straight inside ChatGPT Airtasker launched a plugin that lets ChatGPT users post real-world jobs and accept offers without leaving the chat. It turns the AI into a helpful middle step between you and local tasks. The plugin connects the conversational interface directly to the Airtasker marketplace so users can describe a job, post it, and receive offers inside the same window. Source: Google News

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.