Engineers aren't afraid of AI (unlike my website map)

🧭 THIS WEEK AT AI SECOND ACT

Howdy, suffice to say from the image in this week’s edition that AI doesn’t always get it right. Neither ChatGPT or Google Gemini could draw an accurate Australian map of my life travels. I did not try too hard to get it better, but a few attempts didn’t work. Oh well. So, this week -

• I’ve been doing a a bit of ‘vibe coding’ using Claude Code - very fun!

• AI transformation at work

As I wrote about, I want AI Second Act to focus much less on hype and useless news, even tips/tricks, and more on bringing useful and valuable insights, discussions and playbooks for AI transformation. AI will completely revolutionise our lives including the work place, so we need to get prepared and move with it.

If you’re enjoying this newsletter, please share it! Forward this to a friend/colleague and have them subscribe at aisecondact.com/subscribe. Thanks!

👉 You can also just hit "Reply" and let me know what you want more (or less) of, or use the poll to give feedback.

My goal is to make this as valuable and practical as possible as we navigate the new AI era. 🚀

🧰 AI NEWS + LEARNING

Here are a few things I found recently:

• Claude can now edit files! This is huge and unlocks a lot of capability since files used to be read only for ‘knowledge’.

  • Aside; I dropped Perplexity and Gemini - I only pay for ChatGPT and Claude now. Claude is great though I wish it had higher usage limits.

• Claude Code creator discussing how coding will change forever and in next 6 months. It seems coding is the first truly massively valuable market for AI.

• Both ChatGPT and Claude now have learning modes! —> Intent is to help you learn stuff without giving away the answer straight away. I guess high schoolers might not like it.

Vibe Coding Experiments

At 52, I’m probably not allowed to use the word ‘vibe’. But anyway, it’s pretty fun actually. My background is software, so maybe I have a bit of an advantage, but really, anyone can do it! Read here to get started. There are tonnes of other tools and I’m thinking about launching a new site just on this topic. Very popular also is Cursor or Windsurf.

For now, here’s some thoughts -

• The technology is incredible - you don’t have to know anything! Simply ask it as you go what to do next! ‘maybe we should save this somewhere’ —> ‘oh yes, let’s put it on GitHub as a repository to keep version control’.

• I created an about-me page here I’m still working on including putting it on a real domain area.

• I created several small repositories on GitHub this past couple of weeks, just to try things and mostly to learn something like how to set up a project or Claude Code settings or a SAAS starter template. Maybe some of these are valuable! Who knows! Anyway, here for free for now :).btw, I haven’t coded in 20+ yrs, so this was me simply using natural language to create things. Crazy incredible and cool.

• First tip [of many] - start small, do one thing at a time. Once done and working and you like it, save it/store it perhaps in Github and then close the chat session. Start a new one to start the next thing/feature/change you want to do.

🗺️ FEATURED INSIGHT

Introducing AI at Work

Engineers aren't afraid of AI. They're afraid of bad rollouts and/or generally crap enterprise tools. I am constantly amazed at how bad enterprise tools are. Generally, a super clunky mix of old and legacy patchwork tools that become ‘too hard’ to improve. Next is the set of tools to ‘satisfy process’ but unfortunately, the cost often outweighs the value.

It’s viewed as too hard or too expensive to upgrade or evolve, yet hundreds or thousands of employees waste time every day struggling with the same inefficiencies.

Over the past month, I've had conversations with colleagues about AI adoption. None of them doubted the technology. What worried them was leadership pushing AI in ways that disrupted workflows, ignored risks, or burned out teams.

Every one of them had a story: a botched tool rollout, a dashboard that no one trusted, or a "pilot" that wasn’t actually that useful.

It's a pattern—and if you're leading AI adoption, you need to recognize it.

The 3-Conversation Playbook

When engineers (or any team members) resist AI, it's rarely about the math behind the model. It's about trust, safety, and respect. Three conversations make all the difference:

1. The "Why This Matters" Conversation
Don't start with tech. Start with business pain points and career protection. Show why this AI project helps the company and why it makes their jobs more valuable, not disposable.

2. The "Safe to Try" Conversation
Engineers are pragmatic. They'll test anything if they believe failure won't be punished. Frame pilots as experiments, not mandates. Make it clear: if it breaks, you own the risk, not them.

3. The "We'll Fix It Together" Conversation
AI tools rarely work perfectly the first time. Build credibility by iterating in the open. Invite feedback, admit trade-offs, and actually fix what breaks.

Skip these conversations and you'll create skeptics. Nail them, and you'll create champions.

Add AI to What You Do This Week

Meeting Notes That Don't Suck

1. Before your next team meeting, drop the agenda into ChatGPT: "Draft a notes template with action owners and follow-ups."

2. During the meeting, type key decisions straight into that template

3. After the meeting, ask ChatGPT to polish it into a crisp summary email

Your team will thank you, and you'll cut post-meeting admin time in half.

Weekly AI strategies for operating executives
— Brett

👉 Hit “Reply” and share your experience — I read every one!

_{Picture by xxx on Unsplash.}