Last year, "vibe coding" let anyone describe an app in plain English and have AI build it. Now Google is betting the same thing can work for design. The company just overhauled Stitch, its AI UI tool, into a full design platform with an infinite canvas, voice editing, instant prototyping, and an agent manager that juggles multiple design directions at once. Figma's stock dropped 8% on the news. For promo professionals, this is a direct line to faster, cheaper creative. Imagine describing a client's merch landing page in a sentence and getting a clickable prototype in seconds. Distributors who build pitch decks, mockup sites, or presentation visuals could compress days of design work into a single conversation. The tool also generates DESIGN.md files that port design rules into coding tools, so your brand guidelines actually follow you from concept to build. If you've been outsourcing design or wrestling with Canva templates, Stitch just reset the clock on what "quick turnaround" means.
Adobe just launched Firefly Custom Models in public beta, and the barrier to entry is surprisingly low: upload 10 to 30 reference images, wait 30 minutes to a couple hours, and you have a private AI model trained on your brand's exact visual style. It costs 500 Firefly credits per model, and you retain full ownership. The trained model stays private by default, so there's no risk of competitors accessing your brand's look. This changes the economics of brand consistency entirely. Instead of fighting with generic AI output that never matches a client's aesthetic, you can train a model on their actual product photos, packaging design, and brand language, then generate unlimited on-brand visuals at scale. Adobe also expanded to 30-plus models including Google's Nano Banana 2 and Runway's Gen-4.5, added Quick Cut for turning raw footage into first edits, and is expanding its agentic Project Moonlight beta. For suppliers running product photoshoots or distributors managing visual consistency across dozens of client brands, this is the tool that closes the gap between "AI-generated" and "brand-approved."
Microsoft is reportedly building a new Microsoft 365 tier -- possibly called E7 -- that would bundle security, Copilot, and AI agent management into one plan. The wild part? AI agents would hold identities and email addresses just like human employees. Imagine an AI agent that handles your RFP responses, manages your order follow-ups, and has its own inbox your clients can email. That's the world Microsoft is building toward. Whether you love it or hate it, the "AI as a team member, not a tool" framing changes how every business owner should think about staffing and overhead, particularly in sales where fastest reply wins the day. It's not entirely there yet but it's one to watch closely!
The team building your commonsku features is already deep in the AI toolbox. On the latest skucast, VP of Technology Dileshni Jayasinghe breaks down how Cursor, Claude Code, and CodeRabbit collapsed build timelines, why even non-engineers on the team now use AI to navigate the codebase, and the skubot features about to land on your dashboard. Pull up a chair and geek out with us on this one.
OpenAI revealed it's merging ChatGPT, Codex, and its Atlas browser into a single desktop "superapp." Google made a similar move the same day, consolidating everything into AI Studio. Both companies reached the same conclusion: the era of separate AI apps is ending. Most of us are juggling multiple AI tools for writing, research, coding, and browsing so this consolidation matters. Instead of bouncing between specialized tools, the major platforms are building unified workspaces where you can chat, build, browse, and create in one place. Anthropic has been doing this with its desktop app and Cowork, and it's clearly working since both competitors are now copying the approach. The practical takeaway? Pick your primary platform now and go all-in because these tools are about to get stickier, more integrated, and harder to switch away from.
Anthropic just published what it says is the largest qualitative AI attitudes study ever conducted, using Claude itself as the interviewer to run open-ended conversations with 81,000 users across 159 countries in 70 languages in a single week. The standout finding: people aren't splitting into pro-AI and anti-AI camps. Most are carrying both hope and fear simultaneously. Professional excellence was the top hope, while AI getting things wrong outranked every other concern, beating job anxiety, losing personal agency, and over-reliance. And the fact that Claude conducted 80K in-depth research interviews in a week is itself a proof of concept for AI as a business tool that simply didn't exist a year ago. BTW: The visuals on this are rad.
📊 Anthropic now captures 73% of new enterprise AI spending according to Mizuho, up from a 50/50 split with OpenAI just ten weeks ago, driven by Claude Code and Cowork adoption.
🔍 Google rolled out Gemini Personal Intelligence to all free-tier US users, connecting the AI to Gmail, Photos, and Search for hyper-personalized answers based on individual history.
🧵 Cool story: "We replaced our content team's workflow with Claude projects."
🤖 Google Labs added an agent step to Opal, its visual workflow builder, enabling anyone to create agentic AI workflows where autonomous agents handle decisions and actions within no-code business processes.
📺 How to use AI to watch 60 YouTube channels and read 120 newsletters without watching or reading any of them.
🛰️ Blue Origin filed with the US government to launch over 50,000 satellites for "Project Sunrise," an orbital data center network designed to ease the AI boom's pressure on terrestrial power and compute resources.
🎙️ Want to know how the team behind commonsku actually uses AI every day? This skucast episode with VP of Technology Dileshni Jayasinghe has the answers.