🧠ChatGPT O3-Mini Reveals Its Thoughts

Today we will discuss: 

• Don’t miss this week’s Workflow Wednesday—AI workflows for CRM automation, LinkedIn outreach, and follow-up reminders—now available at only $20/month!

• How ChatGPT o3-mini thinks—OpenAI updates its AI model to reveal more of its reasoning process

• Meta’s explores human-robot collaboration—Meta’s PARTNR program studies AI-powered robots for everyday household tasks

• In other AI news today—Google adds watermarks to magic editor photos, Meta partners with UNESCO on speech AI, Sutskever’s SSI eyes a $20B valuation, and Amazon plans a $100B AI investment

• Saved the best for last—9 must-try AI tools

In this week’s Workflow Wednesday we provided AI workflows for the tasks our readers needed help with, have a question you want answered? Join here

Your Questions, Answered: Connection Edition

Q1: How can I make sure my CRM stays up-to-date without spending hours manually updating it?

A1: With AI-powered automation through Make.com, Airtable.com and Gmail. Set it up once, and your CRM stays current with every conversation you have. Let the system do the work—Read our guide on how to do this here.

Q2: How do I stand out on LinkedIn without sounding like a robot?

A2: Use Writecream.com. Their AI-powered generated messages can still have a human touch but save you time. LinkedIn outreach doesn’t have to be boring or stiff. It’s all about making a real connection - Read about this AI workflow and more here.

Q3: How do I stay on top of my follow-ups without losing track?

A3: Automate your reminders. With FollowUpThen.com, you won’t have to remember to follow up. It’s all done for you - Read about this guide and more here.

ChatGPT’s o3-mini update lets you see how it thinks

Key Points 

• OpenAI updated ChatGPT’s o3-mini to show more of its reasoning process, making its answers easier to follow.

• The model now organizes its thoughts into clearer summaries while filtering out unsafe or overly complex content.

🤓News - OpenAI is making changes to how its latest AI model, o3-mini, explains its reasoning, responding to competition from rivals like Chinese AI company DeepSeek.  

The company has announced that free and paid ChatGPT users will now see an improved “chain of thought” that offers more insight into how the model arrives at its answers. This update applies to all users running o3-mini, including those on premium ChatGPT plans using the “high reasoning” setting.  

👨‍💻What you'll see - OpenAI isn’t revealing every step of o3-mini’s reasoning process, but it says it has “found a balance.” The model will now think freely before organizing its thoughts into a more structured explanation.  

“To improve clarity and safety, we’ve added an additional post-processing step where the model reviews the raw chain of thought, removing any unsafe content and simplifying complex ideas,” an OpenAI spokesperson said. Also, non-English users will get reasoning steps in their native language, making the experience more accessible.  

🌟Why this matters - OpenAI says this update is designed to give people a clearer understanding of how the model thinks, helping users follow its reasoning more easily. Beyond transparency, this change also improves the overall experience—giving users a better sense of when the model is on the right track and when it might be off.

Read the full story here. 

Meta’s latest AI research tackles human-robot collaboration in the home

Key Points 

• Meta’s PARTNR program explores how robots can assist with everyday household tasks like cleaning, cooking, and deliveries.

• The project includes a benchmark of 100,000 household tasks and a dataset for training AI-powered robots.

🤖News - Meta has announced PARTNR, a research program focused on how humans and robots can work together in home environments. The goal? To see how robots might assist with everyday tasks like cooking, cleaning, and even picking up food deliveries.  

🧐What’s their plan? Meta is positioning PARTNR as both a benchmark and a dataset to study human-robot teamwork. The benchmark includes 100,000 household tasks, from tidying up dishes to putting away toys. Alongside that, Meta is releasing a dataset of human demonstrations in simulation, which researchers can use to train AI-powered robots.  

But Meta isn’t just keeping this in the lab. The company has already tested the PARTNR model with Boston Dynamics’ Spot robot and developed a mixed reality interface to visualize the robot’s decision-making process.  

“The potential for innovation in human-robot collaboration is vast,” Meta said, emphasizing that the goal is to see robots as partners, not just tools.  

With continued AI advancements, the idea of general-purpose robots assisting in both homes and factories is becoming more realistic. Meta, which has been expanding its robotics research alongside its broader AI efforts, sees PARTNR as a step toward making that vision a reality.

Read the full story here.