ChatGPT for macOS can now work with your apps

Pivot 5: 5 stories. 5 minutes a day. 5 days a week.

1. ChatGPT for macOS can now work with your apps

ChatGPT for macOS now supports working with apps, allowing users to provide better answers by analyzing content from coding apps. This feature is currently available for ChatGPT Plus and Team users, with access for Enterprise and Edu users in the coming weeks. To work with a compatible app, users must ensure it is running and click the Work With Apps button in the chat bar.

A banner will appear indicating which apps ChatGPT is working with, and when a message is sent, ChatGPT will include content from those apps. This content is saved in the user's chat history and deleted within 30 days unless previously de-identified or disassociated.

Read the full story here

2. Tiger Global-backed InVideo launches GenAI-based video creation

Indian video editing platform InVideo is launching a generative AI-powered video creation feature called InVideo v3.0. Users can use text prompts to generate videos in various styles, including live-action, animated, or anime, and export them in formats suitable for YouTube, Shorts/Reels, and LinkedIn.

The platform uses a pipeline of different models to parse prompts and generate videos. InVideo is adding a paid plan called Generative Plan, which allows users to generate 15 minutes of video for $120 per month.

Read the full story here

3. Tessl worth a reported $750 million after latest $100 million funding to help it build ‘AI native’ software development platform

Tessl, a startup founded by veteran entrepreneur Guy Podjarny, has raised $125 million in two funding rounds. The company raised a $25 million seed round in April and secured an additional $100 million Series A venture capital round, led by Index Ventures.

The latest funding round values the startup at $750 million. Podjarny's vision for Tessl is to use AI to let software designers specify the functions and features of programs in natural language, with AI models taking care of the rest. The company aims to begin offering the product to customers in 2025.

Read the full story here

4. MIT breakthrough could help reach AGI faster

The study explores the effectiveness of test-time training in improving language models' reasoning capabilities. It found that TTT significantly improves performance on ARC tasks, achieving up to 6x improvement in accuracy. 

Applying TTT to an 8B-parameter language model, it achieved 53% accuracy on the ARC's public validation set. The findings suggest that explicit symbolic search is not the only path to improved abstract reasoning in neural language models.

Read the full story here

5. AI platform Krea will now let you train your own image generator by uploading a set of custom images