Good Morning! Here's what I have for you in today's newsletter:
Anthropic opens Claude Duet, a proactive agent that messages you first
Google brings Gemini to Google Maps on CarPlay for hands-free help
Kling opens the Kling 4.0 API, now also live on fal and Higgsfield
Turn Opus 5.5 into your 24/7 personal researcher with cloud sessions
4 new AI tools worth trying today
AI AGENTS
Anthropic opened a research preview of Claude Duet, a standalone app where Claude can check in first and keep working between messages. US Pro and Max subscribers can join the waitlist for access on iOS, Android, and web.
Duet prepares a morning brief, spots emails and calendar items that need attention, and drafts replies for approval. Give it a bigger assignment, like a deck or dashboard, and it works while you’re away.
Like OpenAI’s Dots, Duet runs on its own cloud computer, flags things that need you, and checks before acting. Duet uses Claude Opus 5.5 in a separate app; Dots runs on GPT-6 Astra inside ChatGPT.
Your Claude memory, skills, and apps carry over. Duet supports more than 6,000 connectors, versus 4,000+ apps for Dots, and never directly sees your passwords. Background checks count toward your plan’s usage.
Your morning doesn’t have to start with an inbox scavenger hunt. Duet can pull out what needs a reply and make progress on larger jobs before you check in. For people already using Claude, familiar memory and apps come along, while sending messages or making payments still needs your final OK.
AI TOOLS
Google brought Gemini to Google Maps on Apple CarPlay, letting drivers ask questions and change plans by voice. Tap the Gemini icon on the map’s right side to begin. It’s available in the US, with a global rollout underway.
Ask for a park with a skyline view, check whether it’ll be crowded, and find out if rain might spoil the visit. Gemini can answer and offer to start navigation.
Change your route by voice, ask to avoid tolls, or find somewhere to refuel or park along the way.
Report traffic incidents and road hazards aloud. This is the first time Maps has put a voice assistant directly on the CarPlay screen.
“Let’s stop somewhere else” is easy to say and often fiddly to arrange mid-trip. Gemini lets you find a stop, ask about the destination, and adjust directions in one conversation. Update Google Maps on your iPhone and check for the icon before your next drive.
AI MODELS
Kling opened API access to Kling 4.0, letting developers add its multimodal video model to their own apps and workflows. It’s also available on fal and Higgsfield.
Omni Reference brings characters, objects, and audio into one generation, producing scenes up to 30 seconds long in a single pass. Set start, middle, and end frames to guide how the shot unfolds.
On fal, use Kling 4.0 through its interface or fal Agent, with API access coming soon. fal reports more consistent characters and scenes, plus better handling of complex movement.
Kling’s MCP server and CLI let AI agents generate videos, too. Native 4K and 10-bit HDR output are still on the coming-soon list.
The character looks right, but the voice changes. Fix that, and the next shot looks different. Kling 4.0 aims to cut those repeat attempts by combining visual references, audio, and key frames in one generation. API access brings those controls into your own tools, though projects that need native 4K will have to wait.
HOW TO AI
A fresh research brief every morning, straight from cloud sessions.
Claude Code cloud sessions can check your sources on a schedule: Opus 5.5 reads what's new, skips stories it has covered, and saves a brief for you. You'll need an eligible Claude plan, a GitHub account, and a browser.
Step 1: Set up the desk
Create a private GitHub repository called research-desk with a README. Open claude.ai/code, connect GitHub, and allow access to that repository. In the Default environment's settings, allow the source sites you need through Custom network access, or choose Full.
Step 2: Get your first brief
Start a cloud session and choose Opus 5.5 if available. Use the full guide's prompt to create CLAUDE.md for research rules, sources.md for sources, and seen.md for covered stories. Run the first brief and check its dates and links. Save it in the repository; test Research Desk page publishing if available.
Step 3: Schedule it
Open claude.ai/code/routines and create a task using the daily prompt. Choose your model, repository, environment, and a morning time with an explicit timezone. Remove unnecessary connectors under Advanced settings. Confirm when the next run is due.
Step 4: Test it now
Select Run now and inspect the session in History. Previously covered stories should stay out, and updated files should reach the branch used by future runs. If nothing new happened, the brief should say so. Check that any page update keeps the same link.
Step 5: Keep it useful
Read a few briefs each week and watch your plan usage. Reduce sources or frequency if needed. Keep GitHub connected, and check the task after any connection failure. A successful run status confirms completion, so you still need to judge the research.
P.S. You can access all my AI courses, trainings, prompts library, and $500+ worth of AI tool discounts here.
ElevenLabs launched Chimera in ElevenCreative, combining Eleven v4 voice with ByteDance’s Seedance 2.5 video model to create talking characters. Upload or create a character, write the dialogue, and shape its delivery with tags such as [whispers], [laughs], and [yells].
Anthropic updated Claude Security to keep checking code as it changes. Connect your repositories for an initial deep scan, then Claude scans again as updates arrive. Available in public beta on Claude Enterprise.
xAI gave Grok Bot the ability to create reports, memos, and dashboards as pages. Edit them alongside the Bot, or ask it to keep the information current for you.
🧊 Spline: Launched a ChatGPT and Codex plugin for creating and editing interactive 3D scenes, with the editor open beside your conversation.
🎮 Manus: Added automatic rigging and animation to its game builder, helping 3D characters move without making you animate every action.
🔎 Cohere: released Rerank 5 in Pro and Fast versions, with Context Focus returning only the most relevant passages to send to your LLM.
✂️ Hermes DaVinci Resolve plugin: Lets Hermes Agent assemble script-based rough cuts in Resolve Studio or export timelines you can import into free Resolve.
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
✓ $500+ worth of AI tools