Microsoft Research的动态

AI app screens- Elevate your user experience with sleek AI-powered interfaces that adapt to every touch.

It looks like text in AI image generation is now solved. Attached is from the new?Ideogram 2.0. You can choose "design" mode and get something like this. You can do this on their website, in their new app, or... now in an API! (I don't think they had this before.) See https://ideogram.ai. The text-to-image model improvements just keep on coming.

Flow AI's #VisualProgramming allows users to ?? visually assemble app components, define logic, and ??? map out the flow of interactions through a drag-and-drop interface. Link in the comments.

Why is ??? vision ??? necessary for our AI agent to work properly? With purely text-based LLMs, our agent only had the publicly available DOM code to navigate the app. >> After we introduced multimodal LLMs The “structure” of apps is much easier to comprehend using vision. Often, the DOM does not represent well how objects on a website relate to each other. EXAMPLE: In the screenshot, there is a three-dot menu for a specific list entry ? In the DOM, it is an object with no distinct features, and it becomes hard to derive its meaning. Yet visually, it is very clear that this element indicates further options for the specific document.

Looking to understand how AI is helping brands to improve the customer experience and map the customer journey? ?? Head to the Foodservice stage at 1.30pm to see first hand how we're doing just this! See you there #NRF2025

Ooops, I'm a little late with my #TechTalkTuesday video today! Today I'm demonstrating one of my favorite AI tools for design, #Removebg. This app is great for removing the background of an image so that you can use it for your custom designs. #TechTools #ImageEditing #CreativeSolutions #Innovation #BackgroundRemoval

One of the biggest challenges people face with new technology is learning how to use it effectively. We've dedicated countless hours to designing an intuitive app while continuously seeking ways to improve it. One significant enhancement we've made is the addition of voice control for navigation. This feature enables you to quickly access exactly what you're looking for and also provides the option to manually choose from a list of AI-powered features.

?? NEW Background feature - ???????????????????? ????????????! We’re excited to introduce a brand-new capability: change your image’s resolution directly through the Background page in our web app! We’ve included common resolution options in a simple dropdown menu, and behind the scenes, we’re using AI outpainting technology to enhance your images. This is a great way to increase the resolution of your image if you need it ??? #generativeai #fashn #virtualtryon

This app transforms an Atlanta apartment search into a seamless experience using an LLM to update a Google Map and UI dynamically. It’s the kind of demo that could turn into a product launch in three months—but I’m sharing the entire process now. The app is live here: https://demo.hanakano.com But the real value is in the blueprint: a 4-day, free email course that walks you through: - The planning process, including my storyboards - The development, with full access to the code - The deployment, including the prompts that power the app If you’ve ever wanted to build a conversational AI product, this is your roadmap. Get the emails here: https://lnkd.in/gMdBHPK8

Struggling with #photovoltaic power prediction? ?? ?? Learn how Eaton streamlined the process with low-code #MATLAB apps and deployed the entire pipeline as a user-friendly interactive app at #MATLABEXPO. Build and deploy #AI models faster! https://spr.ly/6049SNDlN