- Everyday AI
- Posts
- OpenAI's Agentic AI: 'Operator' leaked
OpenAI's Agentic AI: 'Operator' leaked
What you need to know about OpenAI's 'Operator', Google now the top AI model, ChatGPT ships updates and more.
Everyday AI
November 14, 2024
š Subscribe Here | š£ Hire Us To Speak | š¤ Partner with Us | š¤ Grow with GenAI
Sup yāall! š
Itās been a quick minute since I was actually LIVE on the livestream after taking a lil anniversary trip with my wife
Speaking of tripsā¦. anyone going to the Microsoftās Ignite conference in Chicago? Should be some major AI announcements.
Iāll be there and will likely be doing a show or three from Ignite. (Still getting those details worked out with organizers.)
Are you going to Microsoft's Ignite 2024 Conference in Chicago?(Vote to see live results) |
Lemme know!
(Letās connect on LinkedIn, and tell me youāre from the newsletter. Ya knowā¦ stranger danger.)
Outsmart The Future
Today in Everyday AI
8 minute read
š Daily Podcast Episode: New details of OpenAIās agentic AI codenamed Operator have just been leaked. What is it? Will it change how we work, or is it just smoke? We answer all. Give it a listen.
šµļøāāļø Fresh Finds: The FTC investigating Microsoft, how NVIDIA became a $3 trillion empire, Sema4 launches AI agents, and more. Read on for Fresh Finds.
š Byte Sized Daily AI News: Googleās new Gemini knocks ChatGPT from the top spot, ChatGPT releases big updates for Mac and Windows, Gemini Live (finally!) comes to iPhones and more. ā read on for Byte Sized News.
š AI In 5: Wanna run a local LLM on your machine? Sweet. Today, we show you how. See it here
š§ Learn & Leveraging AI: How will OpenAIās new āOperatorā AI agent system impact you? We break down the details from the latest ChatGPT leaks. Keep reading for that!
ā©ļø Donāt miss out: Did you miss our last newsletter? We talked about ChatGPTās Canvas mode, AMDās layoffs, Grokās new features, Microsoftās new industry-specific models and a lot more. Check it here!
Looking at the leaks: OpenAIās new agentic āOperatorā š°
Details on ChatGPT's new agent system, Operator, just dropped š
ā³ What will it include?
ā³ When will they be released?
ā³ How is it different than Anthropic or Google's AI agents?
ā³ Should I let these agents take my job?
We'll break it down and tell you what you need to know.
Join the conversation and ask Jordan questions on ChatGPTās reported new āOperatorā agent here.
Also on the pod today:
ā¢ Overview of AI agents as of today š§āš»ļø
ā¢ Reported capabilities of OpenAIās āOperatorā š¤
ā¢ Comparison vs. Anthropic and Google š„
Itāll be worth your 47 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Hereās our favorite AI finds from across the web:
New AI Tool Spotlight ā Zepic is an AI-centric customer engagement platform, Futudo is kinda like an AI-powered horoscope that learns from your preferences and Vocaldo is an AI tool that converts speech to text in more than 100 languages.
AI in health ā Evo can create genomes from scratch, as itās a new model trained on DNA.
AI Agents ā Sema4.ai launched its enterprise AI agent platform for automating complex workflows and increasing operational efficiency. Yay! More agents!
Prompt Engineering ā Claude released a āprompt improverā in its console.
yay everyone sharing stupid prompts online bookmark this.
(then stop sharing copy and paste prompts thanks)
ā Jordan Talks Everyday AI (@EverydayAI_)
7:59 PM ā¢ Nov 14, 2024
Big Tech Teaming Up ā According to reports, Amazon is set to journey into āenemyā territory by offering up its chatbot Q to work with Microsoft 365.
NVIDIA ā How did Jensen Huang build NVIDIA into a $3 trillion empire? Barrons has an in-depth look. (Itās prolly the leather jacket, TBH.)
AI at the White House ā What will Trump mean for AI policy in the future? Brookings takes a look.
Big Tech Investigations ā The FTC is reportedly looking into Microsoftās Cloud business.
1. Google takes crown for top-ranked model in the world š
In a stunning leap, Google DeepMind's latest model, Gemini Exp 1114, has soared to the top of Chatbot Arena's rankings. The model clinched the joint #1 spot, surpassing its predecessors in math, hard prompts, and creative writing.
This breakthrough positions Gemini Exp 1114 as a leader in technical and creative AI capabilities, boasting a 32k input context length that enhances its "thinking" ability.
And like that.... @GoogleAI quietly ships the top model in the world.
Dang.
@OfficialLoganK -- any word on when/if this will be shipped to the Gemini front-end?
ā Jordan Talks Everyday AI (@EverydayAI_)
7:25 PM ā¢ Nov 14, 2024
Right now, the model is available for developer and the front-end Gemini chat is still running an older version of Gemini.
Weāll update yāall when the front-end of Gemini chat gets updated!
2. ChatGPT Teams Up with macOS Apps! š±
OpenAI has unveiled a new feature allowing ChatGPT to work directly with macOS coding apps, including VS Code and Xcode, announced on Thursday. While it doesn't yet write code directly into these apps, it bypasses the tedious copy-paste routine by integrating code context with your prompts.
This marks a significant step towards more intuitive AI tools, as OpenAI begins with coding apps due to the surge in popularity of AI coding assistants. Expect broader app compatibility soon, as this feature rolls out to Plus and Teams users and promises future expansions into writing applications.
To enable this, update your ChatGPT for Mac app then enable the setting here.
3. Windows users rejoice: ChatGPT desktop now available for free users š
Speaking of ChatGPT news, OpenAI's ChatGPT app is now free for everyone on Windows, removing the previous paywall that limited access to subscribers.
With a quick Alt + Space shortcut, users can launch the app instantly, making it easier than ever to integrate AI into daily tasks. This move not only democratizes access to ChatGPT but also enhances productivity by offering a seamless, faster experience for everyone.
4. Google's Gemini App Lands on iPhone with Gemini Live Voice š£ļø
Google has quietly launched its Gemini app for iPhone, offering users a new way to interact with AI through text, voice, or camera. The app features Gemini Live, an interactive chat mode similar to ChatGPT's voice function, marking its first availability on iOS.
While the app can't change phone settings or access non-Google apps, it can seamlessly integrate with Google services like Maps and YouTube Music. This move underscores the tech giant's strategy to dominate homescreens and foster user engagement with fast, easy access to their AI chatbot.
5. Report: AI's Energy Crunch could cause power struggles by 2025 ā”
As AI data centers multiply, their electricity demands are soaring, potentially reaching 500 terawatt-hours by 2027, says a new report from Gartner.
This surge is fueling concerns over power shortages, with Gartner predicting that 40% of these centers might not meet energy requirements by 2025. Tech titans like Elon Musk and Microsoft are eyeing nuclear power to bridge the gap, with Microsoft even considering reviving Three Mile Island.
The increasing energy consumption could lead to higher electricity costs and carbon emissions, complicating efforts to combat climate change.
Local LLMs? How to install Metaās Llama on your local machine
You can work offline with a LLM?
Yup.
Today, we show you a quick and easy way you can run the latest open source (or open weights) models like Metaās Llama on your computer.
(Side note ā I used this recently on my flight to Australia where the Wifi wasnāt working! lol)
Check out today's AI in 5.
š¦¾How You Can Leverage:
Your AI assistant is about to grow hands.
Wellā¦ kind of.
Internal meetings on Wednesday revealed their new AI system ā codenamed "Operator" according to Bloomberg reports ā won't just chat with you. It will take control of a virtual computer and execute complex tasks on your behalf.
This isn't another ChatGPT update.
The stakes just changed entirely.
While most companies are still figuring out how to write better prompts, OpenAI is about to let AI agents loose in virtual environments. First through their API. Then everywhere.
The timing isn't random.
Microsoft releases their autonomous agents on November 19th. (Jordanās birfday yaaayyy.)
Anthropic's computer use capability is already reshaping how their enterprise clients work.
Welcome to the age of AI agency.
Hereās what ya need to know about OpenAIās reported agentic system, codenamed Operator. š
1 ā Why Operator Changes Everything š¼
What will OpenAIās āOperatorā agent reportedly do?
Use a computer to perform tasks for users, including activities like writing code and booking travel.
Function as a general-purpose tool that executes tasks in a web browser or virtual environment.
Operate with minimal human supervision and could handle multi-step processes autonomously.
Integrate with existing tools and workflows to simplify complex tasks, such as managing schedules and finances.
Does the above list kinda sound like a big chunk of your job?
Yeah, yeah, yeah. We know.
Does this really have to be about Agentic AI taking jobs?
Kinda, yeah.
A Harvard Business Review study just confirmed the obvious: AI isn't creating more jobs than it eliminates. ChatGPT already decimated demand for basic writing and coding roles.
But the AI chatbot known as ChatGPT was just practice.
Operator marks the shift from AI that suggests to AI that executes. It can navigate systems, access tools, and complete entire workflows while you grab coffee.
OpenAI spent 18 months planning this.
Every ChatGPT interaction trained a bigger vision. Those suspiciously affordable enterprise plans? You weren't just getting a deal. You were training a future AI system that could do your job in the future.
(For better or worse.)
Try this:
Pick your three most repetitive workflows that take 15-45 minutes to complete. Document every step, every tool, every decision point.
Then, start mapping out which of your tasks could, in theory, be completed by an AI agent like Computer Use, Operator or Copilot Studio.
Create supervision checkpoints where humans must verify results. You'll need these when Operator arrives.
Oh, and go read that HBR study here. The results areā¦. not shocking at all. (We told yāall about this years ago, and reminded you recently.)
2 ā The Real Reason Tech Giants Are Moving Now š¦£
Four major AI agents launching within weeks of each other aint a coincidence.
It's coordination.
Microsoft's agents will have native access to your entire 365 suite.
Google's Jarvis will control your Chrome browser.
Claudeās computer use will
OpenAI's Operator gets a virtual sandbox. Different approaches, same goal: rewiring how work happens.
The cloud revolution took years.
The agent revolution will take months.
Once the catās outta bag, thereās no getting back in. All major tech companies are going all in VERY quickly on agents.
Try this:
Weāve been tackling AI agents a toooooooon lately.
Like we told you in 2023, we knew the end of 2024 would be a boon for agentic AI.
Make sure to listen to this interview to get a deep (and wide!) understanding of the larger impact of everything from multi-agent environments to safeguards for AI agents.
3 ā Your New Job Title: AI Pilot š§āāļø
Modern airline pilots don't fly planes ā they monitor autonomous systems and intervene at critical moments.
That's your future.
Sorry. Not trying to be rude or anything.
Thatās just the reality.
LLM and GenAI will soon be running by default on every device you own. And like we talked about above, āAI chatbotsā are just precursors and data harvesters for AI agents.
These AI agents flip the traditional workflow.
Humans become supervisors, setting parameters and handling exceptions. The machines do everything else.
Try this:
Go check out Claudeās Computer Use and use it now.
A few disclaimers ā itās not very good.
Yet.
But Anthropic shipped out a capable (although kinda frustrating) live AI agent powered by their new Claude 3.5 Sonnet.
We gave even non-technical people a step-by-step tutorial on how to install and use Computer Use here so you can get a taste of working with AI agents before they go Main Street Mainstream.
ā
Numbers to watch
42%
42 percent of IT leaders said they believe that if they could integrate AI, it would make the most difference to their organizations
Reply