In partnership with
TL;DR
Claude rolls out enterprise upgrades to Cowork + private plugin marketplaces
Adobe adds AI auto-editing, soundtracks, and voiceovers inside Creative Cloud
Amazon’s AGI lab head exits amid growing AI pressure
Anthropic accuses Chinese labs of using Claude outputs to train rival models
How to prototype real product behavior (not just UI) with Figma Make
A new LLM-powered social network turns GitHub activity into content
Hiring in 8 countries shouldn't require 8 different processes
This guide from Deel breaks down how to build one global hiring system. You’ll learn about assessment frameworks that scale, how to do headcount planning across regions, and even intake processes that work everywhere. As HR pros know, hiring in one country is hard enough. So let this free global hiring guide give you the tools you need to avoid global hiring headaches.
ANTHROPIC KILLS ANOTHER SECTOR?
Anthropic launched major upgrades to Cowork and plugins—letting companies build private AI marketplaces inside Claude. Admins can now create, manage, and distribute custom plugins across teams, with tighter control over connectors, skills, and data access.
Claude can also orchestrate work across tools like Excel and PowerPoint, passing context between apps and executing workflows end-to-end.
What you should know:
Claude’s new “Customize” hub lets admins build plugins from templates or scratch, manage connectors (via MCP), and deploy private marketplaces across the company. More departments are supported out of the box, and enterprise controls are tightening.
Why it matters:
AI is becoming the layer that coordinates work assistance
Enterprises can now spin up role-specific “agents” for marketing, finance, ops, and more
This moves Claude from chatbot → internal automation platform
However…
As Claude gets better at replacing internal workflows, people are starting to panic.
When you let companies replace human execution with AI agents at scale, sure you’re cutting costs, but you’re also reshaping who earns income in the system.
There’s a growing argument online:
If businesses stop paying humans to produce work, those same humans lose purchasing power. And when purchasing power drops, demand drops. Lower demand means lower revenue. The savings you made on labor could come back as lost customers.
It’s the classic economic loop:
Workers aren’t just costs. They’re the buyers.
OUR TAKE
If an AI can handle chunks of ops, marketing workflows, reporting, internal tools… a lot of small SaaS products start looking very replaceable. Some job functions start looking… compressible.
Enterprises don’t just chase cheap.
They chase control.
Compliance.
Stability.
AI won’t wipe out teams overnight. It will just quietly let fewer people do more.
Claude isn’t loudly killing SaaS.
It’s just casually swallowing pieces of it while everyone’s online still debating the ethics.
QUICK HITS
Adobe brings AI editing to everyday video workflows: The company just launched new Firefly tools like Quick Cut that can turn raw footage into rough edits, add soundtracks, and generate voice-overs from simple text prompts.
Amazon’s AGI lab leader is leaving: David Luan, who led Amazon’s San Francisco AI lab and worked on the Nova Act browser agent, is stepping down after less than two years. His exit comes as Amazon faces criticism that its AI efforts are lagging competitors.
Anthropic says rivals used Claude to train their own AI: The company claims DeepSeek, Moonshot, and MiniMax created thousands of fake accounts and ran millions of prompts through Claude to copy its reasoning, coding, and tool-use abilities—using a technique called “distillation” to speed up their own model development while potentially bypassing safety safeguards and export controls.
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
HOW TO PROTOTYPE REAL PRODUCT BEHAVIOR
Figma Make turns static Figma designs into interactive, testable prototypes
⏱️ 5–15 minutes | 🔧 Needs: a structured Figma file (frames + components), basic design system, a clear flow you want to test (onboarding, feed, settings, etc.)
Why build this? Static mockups answer what it looks like. They don’t answer how it behaves. That gap is where teams over-debate edge cases, misalign on assumptions, and ship the wrong flow. Figma Make compresses idea → interaction so you can pressure test product decisions before engineering time is committed.
Steps:
Start with a real flow (don’t generate from scratch):
Design your onboarding, feed, or settings screens in Figma using real components from your design system. Make works best when structure and hierarchy are already defined.Select the frame → Send to Make:
Highlight the flow you want to prototype. Open Figma Make from within the file. You’re enhancing behavior, not redesigning the UI.Give behavioral instructions, not visual ones:
Prompt with logic, not aesthetics. Example:
“Turn this onboarding flow into an interactive prototype. Add role selection, conditional steps, and an empty state after completion.”
Be explicit about conditions, branches, and outcomes.Generate the interactive prototype:
In seconds, Make converts static screens into a clickable experience. You can tap through steps, trigger state changes, and see conditional flows in action, all within your existing design system.Add real-world logic to key surfaces:
Take ambiguous screens (feeds, suggested follows, modals) and instruct Make to add tap interactions, follow/dismiss logic, collapsible cards, etc.
This is where behavioral questions get resolved before coding.Test edge cases intentionally:
Ask questions like:What happens if a user skips this?
What if they ignore it three times?
Does “Follow All” feel pushy?
Interact with the prototype instead of debating hypotheticals.
Share for async feedback:
Send the interactive prototype link to team members. Replace opinion-based Slack threads with something people can actually click.
Expected outcome:
A realistic, clickable prototype that answers behavioral questions early—so founders, PMs, and designers align on interaction logic before engineering writes a single line of code.
Full tutorial:
TOOL OF THE DAY
An LLM-powered social network for developers.
Devs turns your GitHub activity into readable, shareable posts automatically. Connect your GitHub, push code as usual, and the platform uses AI to transform your commits into developer-friendly updates people actually want to read.
It’s like “build in public” without writing threads—your code ships, and your story ships with it.
