Presented by
Good morning, AI enthusiasts. Vibe coding changed how developers build software. Now Google is betting the same shift can happen in design.
With five major upgrades to Stitch, Google is positioning "vibe design" as a new creative workflow where AI handles the iteration and humans set the direction. Will a chat box finally replace the wireframe?
In today's recap:
Google's Stitch evolves into a vibe design platform
Microsoft splits Copilot and model development leadership
Anthropic publishes world's largest AI sentiment study
4 new AI tools, prompts, and more
Google transformed Stitch into a vibe design platform
Recaply: Google just unveiled a complete redesign of Stitch, positioning it as a "vibe design" partner with an AI-native infinite canvas, voice mode, and instant prototype generation.
Key details:
The new AI-native canvas accepts images, code, and product requirements as creative inputs, with a new Agent Manager letting designers work on multiple ideas in parallel across the same session.
The redesign ships 5 major upgrades, including a DESIGN.md system for consistent design patterns.
Voice mode is available in Preview, allowing Stitch to "see" what the user is working on and provide real-time design critiques.
The new canvas, smarter design agent, and instant prototypes are rolling out now, with voice mode in preview and Design Systems available to all users.
Why it matters: Vibe coding took off because AI removed the barrier between an idea and working code. Google is betting the same shift can happen in design. Stitch now turns natural language into high-fidelity UI prototypes, helping designers iterate at the speed of thought rather than drag-and-drop. With voice input for real-time critiques and a DESIGN.md system for design consistency, this is less a tool upgrade and more a new kind of design workflow.
PRESENTED BY MINTLIFY
Your Docs Deserve Better Than You
Hate writing docs? Same.
Mintlify built something clever: swap "github.com" with "mintlify.com" in any public repo URL and get a fully structured, branded documentation site.
Under the hood, AI agents study your codebase before writing a single word. They scrape your README, pull brand colors, analyze your API surface, and build a structural plan first. The result? Docs that actually make sense, not the rambling, contradictory mess most AI generators spit out.
Parallel subagents then write each section simultaneously, slashing generation time nearly in half. A final validation sweep catches broken links and loose ends before you ever see it.
What used to take weeks of painful blank-page staring is now a few minutes of editing something that already exists.
Try it on any open-source project you love. You might be surprised how close to ready it already is.
MICROSOFT
Microsoft hands Copilot leadership to former Snap executive
Getty Images
Recaply: Microsoft just reorganized its Copilot AI leadership, placing ex-Snap executive Jacob Andreou in charge of consumer and commercial experience while freeing Mustafa Suleyman to focus entirely on frontier model development.
Key details:
Andreou takes on the role of EVP of Copilot, unifying consumer and commercial product under one org spanning experience, platform, Microsoft 365 apps, and AI models, reporting directly to CEO Satya Nadella.
Microsoft's Copilot had 6 million daily active users in February, compared to ChatGPT's 440 million and Gemini's 82 million, according to Sensor Tower data shared with CNBC.
Suleyman said the shift reflects his belief that "the model is the product," committing to build cost-optimized enterprise model lineages over the next 3 to 5 years, he told CNBC.
Andreou and the leadership team will spend the next few weeks aligning teams, with Suleyman remaining involved in day-to-day Microsoft AI operations despite the narrowed focus.
Why it matters: Copilot has just 6 million daily users compared to ChatGPT's 440 million. That gap tells the whole story. Microsoft brought in a growth exec from Snap to fix the product's engagement problem, while freeing Suleyman to focus on building better models. It's a bet that the product challenge and the model challenge need different leaders. Whether Andreou can close the gap remains the bigger question, but Microsoft is at least naming it.
ANTHROPIC
Anthropic runs world's largest AI user study
Anthropic
Recaply: Anthropic just published findings from the world's largest qualitative study of AI users, interviewing 80,508 people across 159 countries and 70 languages on their hopes and fears about the technology.
Key details:
The study used Anthropic Interviewer, a version of Claude prompted to conduct conversational interviews, allowing open-ended responses to be collected, categorized, and analyzed at a scale previously not possible.
80,508 people responded in one week, with 67% viewing AI positively globally, though optimism runs higher in South America, Africa, and Asia than in Europe or the United States.
Economic concern was the strongest predictor of overall AI sentiment, with people who benefited most from AI also most likely to fear what it could cost them, according to the findings.
Interviews were conducted in December 2025 and published on March 18, 2026, with Anthropic saying it plans to run the interviewer regularly on different topics going forward.
Why it matters: Most AI research uses surveys with fixed answers. This study collected open-ended interviews at a scale that wasn't possible before AI. The 67% positive number is notable, but the real finding is simpler: fear and hope don't split people into separate camps. People who benefit most from AI are also the most worried about its costs. That tension is a more honest starting point than either pure optimism or pure skepticism.
TOOLS
Trending AI Tools
🎨 Stitch - Google's AI-native design canvas for creating, iterating, and prototyping high-fidelity UI from natural language
🤖 MiniMax Agent - MiniMax's general-purpose AI agent, now powered by the new MiniMax-M2.7 model
💻 Manus Computer - Manus's AI agent now available as a desktop app, working directly with local files, apps, and terminal commands on your computer
🚀 CMO - Okara’s agent-powered marketing suite for running campaigns, content, and growth strategies
NEWS
What Matters in AI Right Now?
Logan Kilpatrick revealed that Google AI Studio is getting a completely rebuilt vibe coding experience today, with the team spending 4 months rebuilding it from scratch to help users bring ideas to life.
Perplexity launched its Comet AI browser on iOS, with the app now available on the App Store.
Perceptis AI introduced a presentation generation platform that turns a prompt into a board-ready, editable PowerPoint deck grounded in real research and structured like a consulting deliverable.
OpenAI released GPT-5.4 mini and GPT-5.4 nano in the API, Codex, and ChatGPT, with mini running more than 2x faster than GPT-5 mini and approaching GPT-5.4 performance on SWE-Bench Pro at 54.4%.
Mistral AI introduced Forge, a system for enterprises to train frontier-grade AI models on proprietary data, with early partners including Ericsson, ASML, and the European Space Agency.
The Pentagon is building alternatives to Anthropic's AI after the two sides failed to agree on access terms, with the Defense Department's chief digital officer saying engineering work has begun and alternatives will be ready "very soon."
Hollywood is embracing AI as production infrastructure rather than a replacement. Amazon MGM Studios is already using it to map scenes before filming and to help teams see "near-final visuals" the same day as shooting.
Microsoft is considering suing Amazon and OpenAI over their $50 billion Frontier AI deal, arguing that it violates a key clause requiring OpenAI's models to run through Azure.
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
