AI News – May 1
Claude goes full agent mode Higgsfield keep on shipping Gemini’s getting image editing Suno releases version 4.5 Krea lets you paint prompt and once again the robots remind us who’s boss. Here’s today’s AI news. Claude have added a new integrations feature that allows users to connect Claude to services like Jira, Zapier, Intercom and many more.
[00:20.8]
This of course uses Anthropic’s very own MCP protocol to make these connections and and it opens up an incredible array of possibilities to let the AI take advantage of other tools to do agentic style work. What’s more, Claude’s Research mode upgrade allows it to search the web, access Google Workspace and these newly integrated apps for up to 45 minutes per session.
[00:42.7]
This might be the most comprehensive system we’ve seen so far for unlocking the power of an AI model beyond its own capabilities. Right now it’s only for the top tier Claude subscriptions, but it should be coming to the pro plan soon. The Higgsfield video generation model has been one of the breakout stars of 2025.
[00:59.5]
Part of their rise to the top has been due to their ability to seemingly ship a new feature every few days. Today they continued that trend by introducing start and end frames. This allows users to define the first and last frames of the video clip for greater control. This is a feature that is already present in many other models, so adding this will keep Higgsfield in a position where they can compete for the top spot.
[01:22.6]
Another company that just can’t stop shipping updates is Google. Today they announced that they are testing a new set of image editing features in Gemini. Like changing backgrounds, swapping objects and more. The Gemini offering just keeps getting better. SUNO has been the leader in AI music creation for quite a while now, and they’ve just dropped version 4.5 for Pro and Premier subscribers.
[01:44.8]
This update boasts a wider range of genres, richer vocals and enhanced prompt understanding. Promising to make this their most sophisticated version to date Krea have introduced an amazing new feature called GPT Paint. This allows users to visually prompt ChatGPT using drawing marks, shapes, notes and reference images to enhance the level of creation control.
[02:07.4]
Taking more direct control of the image composition in this way feels vastly superior to the typical text prompt. And finally, our weekly reminder to be kind to the robots. A video was posted online of a robot malfunctioning while it was being programmed.
[02:22.7]
In a terrifying display of high energy flailing, the robot went berserk and did its best to take out its owners. If we do end up in a Skynet scenario, we’ll all know where it started.