Good Morning! Here's what I have for you in today's newsletter:
Suno introduces Suno v6, their most powerful model yet
Google Pics rolls out to Workspace and AI Pro and Ultra subscribers
Grok can now trade and manage a Coinbase portfolio directly from chat
Four real changes to get 100% out of GPT-6 Astra, confirmed by OpenAI
4 new AI tools worth trying today
AI MODELS
Suno launched v6, its most powerful generation yet, built in partnership with artists and the music industry across three new models, V6 as the flagship, V6 Wild for exploration, and V6 Mini for free users.
A single request can now pull elements from multiple songs at once, taking vocals from one track and drums from another while adding new lyrics and a different musical style, all in one instruction instead of separate steps.
Small, targeted edits work without touching the rest of a song, changing a single word throughout a track or making the chorus more energetic while leaving the verses untouched.
Max mode gives the model more compute before generating, reducing drift on longer songs and keeping the musical style consistent across a track past two minutes.
All changes point at the same thing, less setup work standing between an idea and a finished song. Pulling elements from separate tracks, fixing one word, and holding a consistent style over two minutes were each their own separate struggle before.
V6 Wild exists to surprise you rather than nail exactly what you asked for, and V6 Mini brings that same jump in quality to anyone on the free tier, not just paying users.
AI TOOLS
Google rolled out Google Pics, its image creation and editing tool inside Google Workspace, to AI Pro and Ultra subscribers and most Workspace business customers, built on the Nano Banana image generation and editing model.
Object segmentation lets a user isolate a single element and edit it in place, requesting targeted changes through text comments and applying several edits at once.
Text inside an image can be edited or translated directly without disrupting the surrounding design or changing the font.
Integration with Workspace starts in Docs and Slides today, with Drive support coming in the following weeks.
Object segmentation isolates the element being edited from everything around it, a different mechanism from ChatGPT Images 2.5's comment-based approach, though both aim at editing one part without touching the rest.
Text editing that preserves the original font solves a specific, recurring headache in translated marketing materials, where a redone image often looks visibly different from the source.
AI PLUGINS
Grok added the ability to trade and manage a portfolio on Coinbase directly from chat, letting users check balances, analyze data, and place or cancel trades for any available asset through the Coinbase connector.
Balances and trade data can be checked directly in chat, without opening the Coinbase app or website separately.
Trades can be placed or canceled on any available asset, covering the full range Coinbase supports, not just a limited set.
The Coinbase connector handles the actual execution, so Grok acts as the interface while trades run through an existing, regulated exchange.
Checking a balance and placing a trade used to mean two separate apps and two separate logins. Now both happen in the same chat window, cutting out the app-switching that usually comes with managing a portfolio.
HOW TO AI
Old habits backfire on this model. Tell it to always ask before doing anything, and it stops right where you wanted it to keep going. OpenAI's own guidance agrees on the same shift, Astra needs less spelled out, not more.
Step 1: Tell it when to just keep going
Give it a task with real gaps and it's more likely than older models to stop and ask, even when the missing detail barely matters.
Keep going and make reasonable assumptions when a missing detail isn't critical. Just tell me what you assumed. Only stop and ask me if the missing information would seriously change the direction, or if it's something you can't undo.Step 2: Delete old custom instructions
Instructions written for an older model can actively hurt this one. A rule that says "always double-check everything" slows down every answer for no real benefit.
Here are my current custom instructions: [paste what's in your Settings > Personalization > Custom Instructions]. Tell me which of these were probably added to make up for an older model's weak spots, ones GPT-6 Astra likely doesn't need anymore.
Step 3: Make sure model effort is well-spent
A typo fix doesn't need the same scrutiny as something touching money or personal data.
This is a small, low-stakes request. Give me a quick, direct answer, and don't triple-check it or hedge it with a bunch of caveats unless something genuinely seems risky.Step 4: Control the output’s length and format
Left alone, it leans toward long, heavily formatted answers even when a plain paragraph was wanted.
Write this in plain paragraphs, not bullet points or headers. Keep it short and to the point.Step 5: Define what "done" actually means
It can stop to check in while there's still real work left.
Take this all the way through, don't stop partway to check in with me. Only stop early if you genuinely need something from me to continue, or you're about to do something that can't be undone.
P.S. You can access all the AI trainings (including the full version of this one), prompts and workflows if you upgrade.
Firecrawl introduced pay-as-you-go for paid plans, with automatic $5 top-ups the moment usage runs out, a monthly spending cap under user control, and the option to buy credits on demand.
Unity brought itself directly into Claude Code with an official, first-party plugin, shipping curated skills for game development, monetization, and performance optimization, made and maintained by Unity Technologies.
Notion updated AI Meeting Notes to draft an email recap with attendees already filled in from a connected calendar, ready to review and send without leaving Notion.
📱 Kimi Work: Remote Control is now live, leave it running on your computer and keep things moving from your phone.
🔋 Desert Ant Labs: launching with 18 models across audio, vision, and text, SDKs for Swift, Kotlin, and JavaScript, nothing leaves the device.
🔌 Perplexity: API MCP server now supports OAuth, sign in with a Perplexity account for real-time web search and deep research.
📸 fal: multi-angle is now live for H3 Max, set a camera angle in degrees and render the same moment from a new viewpoint in under 3 seconds.
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
✓ Full archive of premium guides with ready-to-use prompts
✓ Structured AI courses (step-by-step, start-to-finish)
✓ Every upcoming premium tutorial