OpenAI is setting the stage for an agentic AI future, making ChatGPT more capable and seamlessly integrated into users’ daily workflow.
On the 11th day of ‘12 Days of OpenAI, ’ the company showcased advancements in ChatGPT’s desktop capabilities, highlighting the evolution of ChatGPT from a conversational assistant to an agentic tool—ready to take on tasks and collaborate seamlessly across your desktop environment.
Today, the company announced support for apps like Apple Notes, Notion, Warp, Xcode, and 30 more. In a demo, OpenAI also demonstrated how ChatGPT can work with the Notion app, fetch content within a selected section of the document, and generate additional, relevant content.
The demo showed how ChatGPT helped analyse Git comments in Warp. The results were visualised in a holiday-themed bar graph. In another demo, ChatGPT worked with macOS accessibility APIs in Xcode.
Moreover, OpenAI announced the integration of Advance Voice Mode with this feature. This lets ChatGPT work with apps based on your voice-based commands.
Finally, the dream demo of talking to ChatGPT about stuff on your computer
Shipping today on macOS:
– Advanced voice works with your apps
– New tools like @warpdotdev
– Extending beyond devs with Apple Notes, @NotionHQ, and more!
Next:
– More apps & capabilities
– Windows pic.twitter.com/xgAMt8TOet— Alexander Embiricos (@embirico) December 19, 2024
This new development comes in the backdrop of OpenAI releasing the ChatGPT desktop version in November this year, marketing its first step into computer vision and agentic control.
Drops Another AGI Easter Egg
But, the demo didn’t compete with another AGI easter egg. In one of the presentation screens, AIM was able to spot a directory in an Xcode project titled “AGI_Interface.swift”.
This isn’t the first time in the last twelve days, and OpenAI dropped another notable easter egg a few days ago, a calendar event titled ‘Super Secret AGI’. This further adds to the expectations that all the announcements of the 12 days could cumulatively give users a sense of general intelligence.
OpenAI also mentioned that a Windows app for ChatGPT will be released soon. However, a bigger revelation is the confirmation of a new agent, who is said to be released in 2025.
“As our models get increasingly powerful, ChatGPT will become more and more agentic,” said OpenAI, and they also added that they will have more to say on it in 2025.
A few weeks ago, reports surfaced that OpenAI is working on a new AI agent titled ‘Operator” – but all the company did in today’s event was just to confirm the plans. Perhaps, OpenAI caved into the heat that it is facing from competitors.
Google recently announced Project Mariner, which can navigate and take action on a web browser tab on your behalf. Similarly, Microsoft announced the Copilot Vision, which can see content on your web browser and provide relevant information. And, of course, there’s Anthropic’s Computer Use, released well before any other tool.
That said, there is just one day remaining in 12 Days of OpenAI, and it looks like they’re saving the best for the last — which is a new and powerful frontier model. It will be interesting to see what OpenAI has in store or how different the model is going to be from o1.
A few benchmarks have been suggesting that o1 is the most powerful AI model yet and even outperforms the Claude 3.5 Sonnet in coding tasks.
O1 BLOWS EVERYONE ELSE AWAY – IT IS A BEAST IN REASONING AND IS ALSO THE BEST IN CODING!!
The new o1 model from 12/17 is #1 on Livebench AI and scores 91.58 in Reasoning!!
Finally, OpenAI also BEATS Sonnet in Coding.
Pretty much the only drawback to o1 is that it takes too… pic.twitter.com/roMDMPckCR— Bindu Reddy (@bindureddy) December 19, 2024
A few days ago, a user on X spotted the GPT-4.5 model with a limited preview capability.
That said, today’s event follows the nature of developer-centric announcements from two days ago. On the ninth day, OpenAI launched the o1 model in the API and upgraded it with function calls, structured outputs, reasoning effort controls, developer messages, and vision inputs.
While OpenAI unveiled other features like Projects, WhatsApp integration, and ChatGPT for Apple Intelligence, the highlights remain the launch of o1 Pro and Sora.
However, Google was quick to respond with Veo 2, and the company’s testing revealed that it outperforms competitors like Kling, Moviegen – and, of course, Sora. The buch doesn’t end there. Just a few hours before today’s event, the tech giant unveiled an advanced reasoning model, Gemini 2.0 Flash Thinking.
The ‘shipmas’ battle is getting quite heated up, and for OpenAI, it all depends on what they’re going to release on the last day of ‘12 Days of OpenAI’ and the final day tomorrow.
The post OpenAI Sets the Stage for Agentic AI with ChatGPT Desktop Apps for Mac and Windows appeared first on Analytics India Magazine.
