AI paints the future & more (April 24, 2025)

Hey friends,

If last week was AI in beta, this week is full send. OpenAI’s image model goes public, Grok gets vision, and Microsoft wants you managing agents like interns. Let’s dive in.

🎨 Image Everything: The AI Visual Playground

OpenAI's API goes full Picasso
The gpt-image-1 model, once reserved for ChatGPT, is now open to all developers. That means anyone building with APIs can now embed high-quality image generation—accurate text, diverse styles, and advanced editing—straight into their apps. Platforms like Canva and Adobe are already on board. Pricing ranges from 2¢–19¢ per image depending on quality, making viral visuals accessible at scale.

Four developers collaborate in a bright office, each working on devices displaying different visuals

Adobe’s Firefly turns up the heat
Adobe’s revamped Firefly app lets users mix-and-match image and video models from Google, OpenAI, and others. It now includes AI tools to mock up visuals in seconds or brainstorm dozens of variations through Firefly Boards.

Grok now sees (and speaks)
xAI’s chatbot Grok got a big upgrade: vision mode + multilingual voice. Point your camera at a doc, road sign, or weird bug—Grok analyzes and explains in real time. And yes, it understands Spanish, Turkish, and Hindi too.

🧠 Work 3.0: Agents, Frontier Firms & New AI Coworkers

AIs are more than tools, they are coworkers

Microsoft drops new AI agents
The tech giant launched “Researcher” and “Analyst” Copilot agents built for deep dives in M365. Their 2025 Work Trend Index shows AI-heavy “Frontier Firms” are thriving—71% say they’re doing great vs 37% globally. These companies are already running hybrid AI-human teams.

Everyone’s a manager now
According to Microsoft’s research lead, AI is no longer just a tool—it’s a teammate. Even early-career workers will soon manage agent squads, with delegation becoming a default job skill. Get ready to be an “agent boss” whether you like it or not.

Claude’s meeting cheat codes
Claude can now scan your Gmail + Calendar to prep for meetings, summarize prior convos, and even suggest talking points. Think of it as a pregame scout for every Zoom you’ve got.

AI assistant scanning email inbox and calendar

⚙️ Tool Time

Guidde – Turn documents into how-to videos in minutes
Genspark – AI slideshows with just a prompt
Reef – Audio-based trend visualizer
Infactory – Boost agent accuracy without prompt engineering

⚡ Quick Hits

Nvidia launched NeMo microservices to help companies build custom AI agents fast
Claude’s safety team had to patch misuse tied to social media bots
Supabase raised $200M at a $2B valuation, with backing from OpenAI’s Chief Product Officer

🧠 Prompt of the Day

Ask history’s greats
“Act as a panel of Leonardo da Vinci, Marie Curie, Socrates, Einstein, Sun Tzu, and Maya Angelou. I’ll share an idea—each gives feedback through their unique lens, ending with a powerful takeaway and one question I should sit with.”

TL;DR

OpenAI’s viral image model goes public via API. Microsoft doubles down on AI agents and shows why “Frontier Firms” are winning. Claude wants to prep your meetings, and Grok just learned to see and speak. AI isn’t just powering tools anymore—it’s reshaping how we work, talk, and create.

That’s a wrap for this week.

Keep your agents close, your prompts clever, and your image credits ready—because AI’s not slowing down.

Catch you in the next one 👋

David