Anthropic has quietly taken a big step forward with Claude. What started as a tool for drafting and summarizing now includes the ability to build and run ​AI-powered apps​ right inside the Claude interface.

This update builds on the earlier Artifacts feature and adds a more interactive, technical layer. Instead of writing specs, waiting for prototypes, or pulling engineers off roadmap items, you can now prototype directly within your AI tool.

Whether it’s a new internal dashboard, a personalized learning experience, or a customer-facing assistant, the path from idea to live demo is becoming drastically shorter and increasingly in your own hands.

Here’s what’s new:

• Build apps within Claude: Users can now create, host, and share interactive AI applications directly in Claude. There’s no separate dev environment, it all happens in the chat window.
• You describe the idea, Claude writes the code: You tell Claude what you want to build, and it generates the code in response. You can then review, adjust, and see the result live. It’s a fast way to prototype without needing to set up tooling.
• Build upon existing artifacts: You can browse the ​library​ and customize existing creations to fit your needs without starting from scratch.
• Artifacts are now interactive: These outputs aren’t just static text. They’re functional apps that can interact with Claude through a built-in API. You can inspect the code, make changes, and share the result.
• Examples from early users
  • ​Small games​ with characters that remember prior interactions
  • ​Adaptive learning tools​ with real-time feedback
  • ​Project dashboard generator​ to create visual status boards
  • ​Writing assistants​ tailored to specific formats or goals
• Simple sharing and deployment: When you’re ready, you can share the app with a link, no deployment required. Claude handles the orchestration, prompt management, and basic error handling.
• Technical flexibility: You can use Claude’s internal API to process inputs and coordinate logic, build rich UIs using React directly within the artifact, fork, customize, and share artifacts across users

Artifacts is available to users of all plans. As of the beta launch, there are some clear limitations: no external API calls yet, no persistent storage, and functionality is limited to Claude’s text-based completion API.

Still, it’s a notable shift. Claude is becoming more than a chat assistant, it’s starting to look like a lightweight, collaborative AI development environment.

Your AI Team: Gemini in Google Sheets, YouTube’s AI Overviews, Google’s “Ask Photo”

Every week, I report on the top updates to your favorite AI tools. This week:

Gemini can now ​generate structured data​ directly in Google Sheets

This might sound simple, but it’s a major step toward practical AI-assisted work, especially for teams that rely on spreadsheets for planning, reporting, or analysis.

Key updates:

• AI in Cells: A new =AI() function allows you to prompt Gemini directly from a spreadsheet cell. You can ask it to generate text, summarize content, or classify entries, all tailored to your selected data range.
• Text Generation, Summarization, Categorization: Gemini can write ad copy, customer summaries, internal emails, or group data (e.g., sentiment, request type) based on context. It supports referencing multiple columns and even custom instructions.
• Context-Aware Output: The model understands cell structure and tone. For instance, it can write in a casual or formal style, or personalize content per customer.
• Refreshable and Selective: Generated cells show when they’re out of sync with updated prompts or data, and you can refresh results with one click. Up to 200 =AI() functions can be processed at once.

By embedding natural language generation directly into cells, Google is moving Sheets closer to a dynamic workspace. Whether you’re building customer summaries, classifying feedback, or speeding up reporting, Gemini is quietly making spreadsheets much smarter.

Quick Hits from your favorite AI tools:

• YouTube adds ​AI Overviews​ to video search. When searching on YouTube, users now see Gemini-powered summaries of video content, helping you find the right clip faster without watching the entire thing.
• “​Ask Photos​” is rolling out in Google Photos. Gemini is now integrated into Google Photos, letting you ask natural language questions like “When did we last go hiking with Sam?” or “Show me the best sunset shots.” It works across your library with contextual understanding.
• Perplexity adds ​real-time financial data​ to Research and Labs. You can now access stock prices, trading volumes, and financial statements directly in your queries, no toggles or plugins required.
• WhatsApp adds ​AI message summaries​. Meta is testing a feature that uses AI to summarize long message threads in WhatsApp, offering a quick catch-up option when rejoining busy group chats.
• Microsoft Copilot now helps you ​create videos from text​. Using Clipchamp and Copilot together, you can turn a prompt like “Make a recap video of our sales kickoff” into a finished video with stock footage, music, and voiceover.
• Microsoft Copilot gets ​extensibility boost​. Admins can now manage agents directly from a new Copilot tab, while developers can build custom agents faster using built-in Office skills and deploy them in Teams and Copilot Chat.

Read more news at the end.