Good morning! Here's what's happening in AI today:
Grok 4.5 is now live everywhere, from grok.com to the X app itself
OpenAI brings voice control to the ChatGPT desktop app
Black Forest Labs unveils FLUX 3, one model for image, video, and audio
How to train Hermes on your workflow so every session starts already knowing you
4 new AI tools worth trying today
AI MODEL
xAI released Grok 4.5, calling it its most capable model yet, and made it available immediately on grok.com, the X app, and both iOS and Android.
The rollout is simultaneous across all platforms, so there's no staggered access or waitlist, you can open the X app right now and start using it.
xAI is positioning 4.5 as a broad capability jump rather than a narrow upgrade, though it hasn't published detailed benchmark comparisons alongside the launch.
This lands the same week as Grok Build workflow upgrades, showing xAI pushing hard on both the consumer model and its developer tooling at once.
A same-day launch across web, X, and mobile means there's no delay between announcement and access. If you use X regularly, Grok 4.5 is already sitting in your app. For a company iterating this fast, it's worth checking in on Grok every few months even if you settled on a different assistant already.
AI VOICE
OpenAI announced that ChatGPT Voice is rolling out globally today on macOS and Windows, letting you control your computer and direct multiple agents in ChatGPT Work or Codex using just your voice.
It's powered by GPT-Live, meaning it can speak, listen, and coordinate work inside the app at the same time, instead of pausing to process each command separately.
The rollout covers Plus, Pro, Business, Edu, and Enterprise plans, so most paid tiers get access on day one rather than a phased release.
This turns voice into a control layer for agents, not just a dictation tool, you can talk through a task while Codex or a work agent executes it in the background.
Voice has mostly been a convenience feature until now. Making it the interface for directing multiple agents changes how you'd actually work through a busy day, especially if you're juggling several tasks and don't want to keep typing instructions.
AI PRODUCTIVITY
Black Forest Labs introduced FLUX 3, a multimodal model built for image, video, audio, and action prediction, with FLUX 3 Video now available in early access.
The model is jointly trained in one unified architecture, meaning the same system handles multiple media types instead of stitching together separate tools for each.
Black Forest Labs is already showing work extending the model to predict actions for robotics, including collaborations with mimic and Audi.
Early access to FLUX 3 Video means you can start testing the video capability now, ahead of a wider rollout.
Most creative AI tools force you to jump between separate apps for images, video, and audio. A single model trained across all of them, with an early nod toward robotics, points at where generative tools are heading next.
HOW TO AI
Hermes has a memory system. Most people never fill it in, then wonder why every session starts from scratch.
Three files hold everything:
MEMORY.md: What your work is, who it's for, what context any task assumes.USER.md: How you work, what formats you prefer, what you always correct.SOUL.md: Voice, tone, what to avoid.
Fill those in before anything else. Twenty minutes. Affects every session after.
Then pick one task you do regularly. Run it in a Hermes session in full detail. A few days later, run it again, and when the output isn't quite right, say so specifically. "Keep this under 200 words next time." It will write that correction directly to memory.
Once you've done the task two or three times with corrections, ask:
Based on how I have asked you to do this task over our last
few conversations, write this up as a reusable skill. Include
the step-by-step procedure, the quality standard I have been
expecting, and common mistakes to avoid.Hermes searches your session history and writes a skill file (a plain markdown document) directly to ~/.hermes/skills/. You can read it, edit it, delete it.
If the first draft is missing something, tell it what specifically. It reads the file, patches it, and confirms what changed. After that, the next time you need the same task, this is all you type:
Research brief: what's new with Grok 4.5 in the past 7 daysHermes will load the skill automatically and deliver a structured brief with sourced developments, an underexplored angle, verification flags, and a word count. Everything the skill specified.
You can access the full guide here
P.S. You can access all my prompts, workflows and AI trainings if you upgrade here
Claude rolled out voice mode on its more capable models, now able to reach tools you've connected mid-conversation and support many more languages.
OpenAI is launching Health in ChatGPT to US users, letting you securely connect Apple Health and supported medical records for more informed conversations.
Google introduced selfie video sign in for your Google Account, so you can get back in even without your password or usual device.
HOT AI TOOLS
🎨 Canvas UI: Open source html-in-canvas component library with 24 components.
🔍 Tavily: Real-time, LLM-optimized web search now built into Grok Build.
🖌️ Hermes Agent: Lets your Hermes agent read and edit your tldraw canvas live.
🤖 OpenArt MCP: Turns one product photo into banners, creatives, and video ads inside
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