Published Date:
July 31, 2025

10X AI Outcomes with ChatGPT o3

We’re barely scratching the surface of what today’s AI can do. Here's how reasoning models like GPT-o3 can finally help you ‘think,’ not just complete prompts.
Daan van Rossum
By
Daan van Rossum
Founder & CEO, FlexOS

Coaching over 1,000 business leaders, it stands out to me that we remain distracted by hype and don’t focus enough on foundational AI knowledge.

Case point: everyone is buzzing about the upcoming ​ChatGPT-5​, but as Ethan Mollick highlighted, people are barely using what ChatGPT can already do today:

In this 10-part series, I want to dive into all of the ways that we’re underutilizing the capabilities of AI as it stands today, starting with Reasoning Models.

What AI tool or feature would you like me to highlight? Reply and I’ll try to include it.

10x Your AI Results, Lesson 1: Welcome to the Reasoning Era (ChatGPT-o3)

Most people using ChatGPT today are stuck interacting with the default 4o model.

This model is great for writing, editing, and answering general knowledge questions.

But it has a key limitation: it doesn’t “think,” pause, reason, or problem-solve. It completes your prompts based on training, not logic.

That’s fine for surface-level tasks. But if you ask for multi-step decisions, deep analysis, or nuanced trade-offs, it can struggle or just ​hallucinate​ solutions.

That’s where ​reasoning models​ like OpenAI’s new o3 come in. (It’s not the only reasoning model, but ChatGPT happens to be the dominant AI platform by miles.)

o3 isn’t just the next version of ChatGPT. Its ​introduction in April 2025​ represented a shift in AI capabilities.

Instead of instantly generating an answer, o3 “thinks aloud” internally, simulating reasoning before speaking. It might research something, run a calculation, or evaluate multiple ideas before choosing what to respond.

This leads to you getting more accurate, insightful, and decision-ready responses, especially on complex or unfamiliar tasks.

Your First Day with o3

If you haven’t tapped o3’s capabilities, then here are a few things to know before starting to use it.

o3 uses “Chain-of-Thought” Reasoning

Ask a tough question, and it will silently break the problem into steps, something ​prompt engineers​ call ​chain of thought​.

This makes it especially useful for planning, coding, math, research, and logic-heavy decisions.

It Can Use Tools Autonomously

o3 can search the web, analyze data files (CSVs, PDFs), interpret images, and even run calculations, without you having to manually direct it to do so.

It’s Multimodal

Like 4o, it’s multimodal. That means your prompt doesn’t have to be text only.

You can upload a chart, contract, photo, or screenshot and o3 will read and reason over it. It’s the only reasoning model with vision built in.

It Takes Its Time (and That’s a Feature)

o3 often spends 30 seconds to 2+ minutes thinking before answering. That’s not a bug. That’s the reasoning process happening before you see anything.

Great First Assignments to Try

Now understanding more of what makes this model so powerful, let’s look at a few ways you can get value from o3 today.

1. Sense-Making

Upload a report or board deck and ask for insights, risks, and actions, tailored to your role, company, or industry.

Marie Fayard, Head of Product Education at DataCamp, ​uploaded​ a 50-page World Economic Forum cybersecurity report and got a 1-page executive summary, “What it means for you” section, and even charts extracted from raw data.

2. Leadership Coaching

Feed in transcripts and ask: What patterns do you see in how I run meetings?

Dan Shipper ​shared​ a week’s worth of meetings and saw o3 flagging a recurring behavior: that he avoided confronting conflict directly.

It became a form of lightweight leadership coaching, with Dan commenting that “because it can reference the exact points in the transcript where this is happening, it helps me sharpen my skills in context—a huge (and hugely expensive) lift for a traditional leadership coach.”

3. Competitive Research

Tell o3 what you’re developing and receive a consulting-style competitive review.

As we're working towards an extension of our popular ​Lead with AI Executive Boot Camp​, I wanted to see what the market for executive AI courses looks like.

I asked o3 to check what it already knew about our existing product and then research similar products, where they differ in their product formulation, and their price points.

o3 generated a helpful table based on fairly extensive research and gave truly insightful recommendations on product development.

4. Strategy Brainstorming

Describe your team or org chart and ask o3 what you’re set up to succeed at and where you’re likely weak.

The Every team fed o3 their “about page” and asked: What’s this team optimized to build? The model answered, but also highlighted what the team wasn’t (yet) capable of:

5. Data Analysis

Upload messy CSVs and ask: What’s driving the change? Or where are we exposed?

Uploading a hypothetical personal investment portfolio, Dries Faems saw o3 thinking through the task and providing him with “surprisingly rich analysis” including a visual summary, categorization of assets, and actionable advice.

PS: I found this use case through o3, very meta!

6. Prospect Research

Still working with a BDR? Get more results in less time with o3 by prompting it to generate a list of prospects for you.

​Lead with AI PRO​ member ​Joe Harris​ shared how he used o3 to first research all public sector organisations and departments in the UK, research their email domains, and finally create an export that he uploaded to their CRM in a format that included all the relevant fields.

This created around 500 accounts and organisations that they could then use for proactive sales campaigns. According to ​Joe​, the thinking time was 20+ minutes, but it “would have taken me 2+ days to do manually.”

Tips for Better o3 Results

These use cases are a great start to tap o3 in your daily work. But regardless of what you task your new reasoning assistant with, keep these tips in mind:

  • Be goal-first. Don’t just say “summarize this,” say why and for whom. (See also: ​SuperPrompting​.)
  • Use threads. o3 builds context, so keep your market analysis or OKR work in one chat.
  • Ask for self-checks. Ask if it’s confident in its answer or to double-check work.
  • Feed it your data. PDFs, spreadsheets, screenshots, transcripts, it can absorb them all.
  • Switch to o3. Started a conversation in 4o but realized later it’s better suited for o3? Toggle between models at the bottom of ChatGPT’s latest response:
  • Switch back. o3 is slow and expensive. If the assignment isn't complex, switch back to 4o or tap into one of the other great ​ChatGPT models​.

The Bottom Line: Reasoning with o3

o3 helps you switch your AI usage from predictable writing and answering to deliberate thinking.

For leaders like yourself, this means you have a real collaborator who understands nuance, can challenge assumptions, and generate novel ideas.

Go and try it today. Have a great use case I should feature? Let me know.

In this series, I’ll break down more high-leverage ways to 10x your work with AI. What topic, tool, or feature would you like to see covered next? Email me and I’ll try to include it.

If you made it this far, reply and tell me what you'd love AI to take over in your daily workflow.

Also, please forward this newsletter to a colleague and ask them to subscribe.

If you have any other questions or feedback, just reply here or inbox me.

See you next week,

Daan van Rossum
By
Daan van Rossum
Founder & CEO, FlexOS