- Everyday AI
- Posts
- ChatGPT’s new Projects Mode: How to use it and 3 useful hacks
ChatGPT’s new Projects Mode: How to use it and 3 useful hacks
ChatGPT Projects overview, Salesforce unveils Agentforce 2.0, OpenAI releases dev API updates, Nvidia announces Jetson Orin Nano Super Developer Kit and more!
Everyday AI
December 17, 2024
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
6 minute read
🎙 Daily Podcast Episode: ChatGPT just released Projects mode. Is it any good? How does it compare to Claude Projects? We dive in and break it all down. Give it a listen.
🕵️♂️ Fresh Finds: Salesforce hires 2,000 people to sell AI, Google updates NotebookLM and Grammarly acquires Coda for AI upgrades. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Salesforce unveils Agentforce 2.0, OpenAI releases dev API updates and Nvidia announces Jetson Orin Nano Super Developer Kit. For that and more, read on for Byte Sized News.
🚀 AI In 5: We’re diving into NotebookLM’s recent updates to show you everything that’s new! See it here
🧠 Learn & Leveraging AI: Wondering what ChatGPT’s new Projects mode is all about? Here’s 3 things you should know. Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We talked about ChatGPT Search becoming publicly available, DeepMind unveiling Veo 2 and Meta joining legal battle against OpenAI's for-profit shift. Check it here!
ChatGPT’s new Projects Mode: How to use it and 3 useful hacks 🗂
Is ChatGPT's new 'Projects' mode a game-changer?
Or just a polished imitation of Projects from Anthropic's Claude?
Welp.... there's a lot more than meets the eye.
We've given ChatGPT's new Projects mode a deep test, to bring you 3 time-saving hacks we figured out.
Join the conversation and ask Jordan questions on ChatGPT here.
Also on the pod today:
• ChatGPT Misconceptions ❌
• Custom GPTs vs. Projects 🤔
• Pros and Cons of ChatGPT Projects ⚖️
It’ll be worth your 47 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – MagicMail is an AI email generator, NewOaks AI creates AI phone calls that convert and SantaCard sends a personalized video message from Santa.
Salesforce – Salesforce’s CEO announced that it will hire 2,00 people to sell AI products.
Google – NotebookLM has officially released its latest updates.
AI in Government – An Amazon-hosted AI tool for UK military recruitment has been found to be a risk of data breach.
AI Startups - Grammarly has acquired startup Coda to turn Grammarly into an AI productivity platform.
Anthropic – Anthropic’s X account was hacked.
AI Chips - Broadcom’s CEO is sounding the alarm on a crucial shift in the AI chip market.
AI in Medical – Dexcom is adding AI reports to its OTC glucose monitor.
1. Salesforce Unveils Agentforce 2.0: A Game-Changer for Digital Labor 💼
Salesforce has launched Agentforce 2.0, enhancing its digital labor platform with a wealth of new AI capabilities and integrations, including deployment in Slack. This latest version introduces a library of pre-built skills, advanced reasoning, and retrieval capabilities designed to empower employees across various departments while streamlining complex tasks.
With companies like Accenture and Indeed already leveraging this technology to boost productivity and personalize service, the implications for businesses looking to enhance their operations are significant.
2. OpenAI’s "Mini Dev Day” for 9th Day of Shipmas 🧑💻️
In a fresh wave of updates, OpenAI has officially launched the o1 model in the API, introducing features like function calling and structured outputs, alongside a new "reasoning effort" parameter that enhances cost efficiency for developers. The Realtime API now boasts WebRTC support and a striking 60% reduction in audio token prices, making it more accessible for developers aiming to create real-time voice applications.
Additionally, the fine-tuning API is getting smarter with Preference Fine-Tuning, allowing for more tailored model responses.
3. Nvidia Unveils Jetson Orin Nano Super Developer Kit 🚀
Nvidia has just launched its Jetson Orin Nano Super Developer Kit, a powerful AI computing solution priced at $249, making advanced AI capabilities more accessible than ever. This new kit boasts a remarkable 70% increase in neural processing performance, hitting 67 TOPS, and 50% more memory bandwidth, ensuring faster operations for AI and robotics projects.
With the same hardware as its predecessor and a significant performance boost thanks to a new power mode, this kit is poised to become an essential tool for developers looking to create innovative applications like chatbots and visual AI agents.
4. Google’s 2025 Prediction: The Year Enterprises Scale AI 🏢
In a new report from Google Cloud, 2025 is being touted as the year businesses finally scale their AI efforts, moving beyond experimental phases. The rise of sophisticated multimodal AI and multi-agent systems promises to revolutionize workflows, enabling everything from intuitive internal searches to autonomous task management.
As enterprises shift gears, Oliver Parker from Google anticipates a future where AI agents work collaboratively, unlocking valuable insights across diverse platforms. This transformation could redefine how professionals access information and streamline operations, marking a significant leap in workplace productivity.
5. YouTube Partners with CAA for AI Celebrity Likeness Control 🧑🎤️
YouTube has joined forces with Creative Artists Agency (CAA) to empower celebrities with control over their digital likenesses, particularly in AI-generated content. This partnership aims to create a responsible AI ecosystem that safeguards artists' rights while exploring new avenues for creative expression.
As the tools roll out to a select group of celebrities and creators, this initiative could redefine how public figures manage their images online, potentially enhancing their ability to monetize their likenesses.
6. OpenAI Passes on Sora API for Now 👋
OpenAI has announced it currently has no plans to offer an API for its video generation model, Sora, despite overwhelming demand that led to service interruptions shortly after launch.
This move could set the company back in the competitive landscape, especially with Google recently launching its Veo 2 API, which boasts impressive quality and is already making waves. OpenAI CEO Sam Altman acknowledged the high demand and promised efforts to accelerate access for users.
NotebookLM Plus Updates – 5 new NotebookLM features from Google
Google just released its new updates to NotebookLM!
We’re diving in to show you what’s new, including a Plus tier, and how the new features work.
Check out today's AI in 5.
In our AI in 5 segment on YouTube, we often create different types of videos. Which are your favorites? |
🦾How You Can Leverage:
There’s two types of people in this world when it comes to your computer desktop.
3 Icons.
Or 393 icons.
For fans of the former, there’s now a new feature in ChatGPT that can help you tidy up your chats and better tap into your company’s knowledge.
A few days ago, ChatGPT unveiled its new Projects feature.
Projects can transform how you'll use ChatGPT.
Not because it helps you organize your chats (it does).
Not because it lets you add files (it does that too).
Or because you can add colored folders. (Yay!)
But because it finally fixes the biggest problem with custom instructions: they’ve been broken since Day 1.
That’s right.
The biggest breakthrough in Projects isn’t necessarily the organizational and hierarchical improvements.
It’s the fact that you can now more easily steer and control groups of ChatGPT chats with custom instructions at the folder level.
Previously, you could only have one set of custom instructions for an entire account, which isn’t very helpful if you’re juggling multiple projects, deliverable types, and also using Chat for your personal life too.
If you want the 101 and 201 of Projects, make sure to watch/listen/read today’s episode.
(And for comparing ChatGPT projects with custom GPTs and Claude’s Projects.)
Buuuuuut for our newsletter recap, we decided to skip the entree from today’s show and go straight to the dessert — 3(ish) hacks to get the most outta ChatGPT Projects.
1 – O1 is Absolutely Available in Projects 👀
OpenAI's website still claims Projects only works with GPT-4o.
That’s wrong.
o1, OpenAI's most advanced reasoning model, runs perfectly in Projects.
No special access required. No complex workarounds needed.
The implications are massive. O1's enhanced reasoning capabilities completely transform how Projects can be used for complex analysis and decision-making.
Try This:
Start a new Project. Look at your model selector. There it is – O1 availability, plain as day.
OpenAI's documentation team might want to update their website.
2 – The File Upload Power Play 📂
Here's where it gets interesting. Upload files at the Project level, every chat in that folder can access them.
But you lose O1 access.
Want both? There's (kinda) a workaround.
Instead of Project-level uploads, add files directly to individual chats within your Project folders.
Downside? New chats in your project won’t have access to those files.
Buuuuuuut you can use o1 and maintain access to all your documents in the other chats.
Try This:
Create your Project first. Skip the Project-level file upload. Start a new chat, select o1, then add your files directly to that chat.
You've just unlocked o1's enhanced reasoning while maintaining full access to your documents, all while being nice and organized.
3 – The Model Cloak for Custom GPT Access 🔓
Everyone says you can't use custom GPTs in Projects. Everyone's almost right.
We discovered a three-step process that makes it possible. Think of it as smuggling your custom GPT's insights into Projects.
(BWHAHAHAH!)
Start in a default GPT-4o chat, not in Projects. Bring in your custom GPT via @mention. Get what you need. Then, switch back to GPT-4o.
Now that entire conversation can be moved and live inside any Project folder.
Yay.
Try This:
Open a standard chat. Use GPT-4o. Send an initial message. @mention your custom GPT. After it responds, send one final message in standard mode. The whole thread can now move into any Project.
The conversation remains active – you can pick up right where you left off.
⌚
Numbers to watch
34 Million
AI startup Basis has raised $34 million for an accounting automation agent.
Reply