Mercedes Cuts 500h/Year, OpenAI o3-pro Launch & Apple Challenges AI

Hello!

Last week, the most-read piece was about what happens when a team gets too big, and today we’re continuing with the theme of productivity and leadership.

Read what to do when your team starts underperforming, why banning AI in interviews isn’t the best solution, and how Mercedes-Benz saved 500 dev hours. Aside from that, I’ve got a fascinating paper Apple prepared on how smart reasoning AI models really are, and how Zapier measures AI proficiency.

Oh, and apparently the Universe didn’t start with a bang after all.

Grab your coffee and enjoy Frictionless!

In the Queue

• OpenAI o3-pro Release

• Examples of AI Agents

• 40+ Expo Apps

• Augmented Coding with Translations

• Sam Altman: Singularity is Here

• Take a 2-Week Vacation to Test Your Team

Reduce Friction

Wardley Maps & Pace Layering for Senior Tech Leads and Engineering Leaders

If you’re juggling competing priorities like I often am, combining Wardley Maps and Pace Layering can bring some much-needed clarity. Wardley Maps help you understand where your tech fits in the market, while Pace Layering shows what needs to move, and what’s better left alone.

9 Questions to Ask When You Start to Notice Underperformance

Have you ever had a team member who’s acting… off? Not yet missing deadlines, but slipping just enough to worry. This article suggests nine questions you can ask to start course-correcting without casting blame.

The 2-Week Vacation Test for Engineers and Managers

I don't think it's a controversial take to say everyone on your team, including you, should be able to take two weeks off without everything falling apart. But if the idea of a longer vacation makes you sweat a little, have a look at this article and see why that might be a sign of deeper issues in your team.

LLMs Are Mirrors of Operator Skill

This article was eye-opening. Banning AI in interviews to stop candidates from cheating prevents you from observing how potential candidates use AI. These tools are here to stay, and watching how someone works with them will tell you more than any whiteboard.

Deepen Your Expertise

How Can Modern Tooling Save Mercedes-Benz.io Engineering Time?

Cars aren’t the only fast thing at Mercedes-Benz anymore. Swapping in a few modern tools helped their team cut build times by 34% and linting by 70%, saving around 500 hours a year. It shows regular audits of your dev stack can pay off without needing a full rewrite.

How AI Agents Work

The best way to learn about AI agents is to look under the hood. Have a look at this thread with examples of agents made for a variety of industries and needs. Yes, even the ones for break-up recovery.

40 Best Expo Examples

If you still think Expo’s only good for prototypes, our list might change your mind. Check 40+ apps made with Expo, from fintech to fitness, which made it to the top of the App Store without native teams behind them.

Augmented Coding Technique: Copy From Simpler Language

What if AI-assisted coding started not with translating from one programming language to another? Kent Beck conducted an experiment where he prompted AI to "copy from simpler language" by feeding a basic implementation of a data structure to an LLM.

AI Corner

The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity

Apple challenges the idea that reasoning AI models can reason at all. Their new research shows that while these language models can handle simple tasks, they fall apart under any complexity and struggle even when handed the correct algorithm.

Launching OpenAI o3-pro

OpenAI’s new o3‑pro model is now available to Pro and Team users, and it’s already looking promising. It reportedly performs better on physics, math, and coding tasks, and is 80% cheaper, though it runs slower than its predecessors.

Measuring AI Proficiency at Zapier

Struggling with defining your team’s AI proficiency? Wade Foster, the CEO at Zapier, shared the company’s framework for measuring how well their employees in different roles and departments use AI. Maybe we could apply it too?

The Gentle Singularity by Sam Altman

According to Sam Altman, we may have hit the point where technological progress starts shaping itself faster than we can steer it. Although we’re not surrounded by robots just yet, the digital superintelligence is closer than we might think.

Just Cool

Big Bang Theory Is Wrong, Claim Scientists

A new theory suggests the cosmos didn’t explode into existence, but has been cycling through dramatic expansions and collapses all along. I don’t know about you, but I find it oddly comforting that even the universe might just be going in circles like the rest of us.

Let’s Stay in Touch! 📨 

Do you have any comments about this newsletter issue or questions you want to ask? Drop me a message or book a meeting.