- Practicaly AI
- Posts
- đź§ AI Starts Thinking in Teams: Copilot, Claude, Manus
đź§ AI Starts Thinking in Teams: Copilot, Claude, Manus
Today in AI: Multiple models, devices, and tools are starting to work together like systems—not just standalone features.
đź‘‹ Hello hello,
Copilot just added a feature where you can run multiple models on the same prompt and compare their answers side by side.
Which means you’re no longer stuck trusting a single response. You can actually see where things line up, where they don’t, and decide what holds up before you use it.
Let’s get into it.
💬 Quick note: We’re building something to help teams actually get good at AI (not just use it). → Get early access here
🔥🔥🔥 Three Highly Curated AI Stories
Critique is a multi-model deep research system. Instead of relying on a single model, it combines outputs from multiple models to generate more complete reports and responses. Think of it as layering perspectives to get closer to a “final answer.”
Council takes this a step further. You can run multiple models on the same prompt simultaneously and compare where they agree or disagree. This gives you visibility into how different models reason—and what each one uniquely contributes.
Why this matters: AI is moving from “give me an answer” to “show me the thinking.” That’s a big shift for anyone doing research, analysis, or decision-making.
Manus just made their desktop app controllable directly from your phone.
You can start tasks, access files, and trigger workflows without touching your computer. It essentially turns your phone into a command center for whatever you’re running on desktop.
Why this matters: AI tools are becoming ambient. You don’t have to sit in front of a screen anymore to get work done. The control layer is starting to move with you
This means Claude can open apps, click through interfaces, and test what it builds—all directly from the command line. It’s currently in research preview for Pro and Max users.
Why this matters: This is a step toward AI actually executing tasks, not just suggesting them. If you’ve ever wanted your code assistant to go beyond writing code and actually run it… this is where things are heading.
🔥🔥 Two Pro AI Tools To Try Today
1. đź§ Claude Cowork Projects Guide
This guide breaks down one of the most important (and underused) features in Claude right now: Projects.
Projects let you persist context—files, instructions, and memory—so Claude doesn’t start from scratch every time. If you’re doing recurring workflows (content, research, analysis), this is where you get real leverage.
This is a wild one. Banana Claude turns Claude Code into a creative director for image generation using Google’s Gemini Nano Banana models.
Instead of just sending prompts, Claude interprets your intent, selects the right “expert lens,” and constructs structured prompts using Google’s official framework. It’s much closer to how a human creative team would approach visuals.
Is this you? Your team is using AI. But they’re not getting better results.
We’re fixing that. Join the waitlist to find out how.
🔥 Things You Should Know About AI
Create Instagram carousels inside Claude (without an image generator)
You don’t actually need Midjourney or DALL·E to create polished carousels anymore.
1. Take a reference carousel and extract its design elements (fonts, layout, spacing, color system).
2. Create a Claude Project and define these as your “design system” in instructions.
3. Feed Claude your content (newsletter, post, etc.).
4. Ask it to generate carousel slides using code-based layouts instead of images.
5. Export the slides as PNG files and post directly.
💬 Quick poll: What’s one AI tool you’ve tried recently that actually stuck?
Did you learn something new? |
Until next time,
Kushank @PracticalyAI


Reply