In this week's newsletter, Aaron Moncur has a conversation with Aaron Eden, AI Process Automation leader at Intuit and co-founder of AI Trailblazers, a nonprofit democratizing access to AI tools through apprenticeship programs.
You basically have the Einstein of every type of job on the planet in your pocket with these tools.
In this episode:
How planning prompts with expensive models (Gemini 3 Pro, Claude 4.5 Opus) then executing with cheaper models saves money without sacrificing quality
Why vibe coding lets engineers build in unfamiliar domains but requires verification checkpoints to catch mistakes
How Lindy's "if ChatGPT and Zapier had a baby" approach enables no-code AI automations with 6,000+ integrations
Why you shouldn't trust AI output despite 80-90% benchmark performance, treating it like a brilliant intern who lacks real-world experience
Bonus Content:
Annual Reset Part 3: Lessons from a 50 Year Old Book About Tennis
S7E3 Aaron Eden | How Engineers Can Use AI Today
Aaron Eden started programming at 12 and encountered neural network papers on bulletin board systems in the mid-80s. The math was over his head then, and he admits it still is today. But that hasn't stopped him from building AI-powered systems that help engineering teams move from T-shaped employees - broad knowledge with one deep expertise - to what he calls "unicorn employees" who can go broad and deep across multiple domains with AI assistance. His latest experiment at Intuit has him playing both business analyst and automation developer simultaneously, a role combination that would have required two people just a year ago. The engineering workflow distinctions that used to define career paths are blurring fast, and Eden's approach to AI tool selection shows exactly why verification matters more than ever.
>If YouTube isn’t your thing, check out this episode and all of our past episodes on Apple, Spotify, and all the rest.
When was the last time you stepped back to think about your career - not just your next task, but your actual trajectory?
Your career is a system. Is it time to redesign it? Join us for our 3-day intensive retreat to reset, recharge and be intentional about your career and life.
Most engineers are running full speed… but they’re not always sure where they’re headed, or why.
They’re skilled, smart, and successful on paper. But inside? Many are feeling stuck, burned out, or disconnected from what once lit them up and perhaps why they became an engineer in the first place.
That’s not a personal failure.
It’s a systems problem. And the system is your life.
We’ve been trained to optimize output, hit deadlines, and keep growing technically. But what about growing intentionally?
Consider asking yourself these questions:
What if your work aligned with your values?
What if your leadership flowed from clarity, not pressure?
What if you actually loved the path you’re on?
Those aren’t idealistic questions. They’re practical ones.
And they’re exactly what is explored at the Intentional Engineer Retreat.
It’s not a conference or a career bootcamp. It’s a 3-day, high-impact reset for engineers and technical leaders who want to reconnect with purpose, rediscover their vision, and design a more fulfilling future.
Think deep reflection, practical frameworks, and authentic connection with others who want to live and lead with intention.
Led by Jeff Perry, award-winning author of The Intentional Engineer, this is an opportunity to take time to design your career.
April 9-11 | Spokane, WA
If you’ve been craving space to think, breathe, and be again, this might be your sign.
Curious? More info and sign up at the link below!
The Two-Model Strategy: Expensive Planning, Cheap Execution
When Eden builds software with AI assistance, he doesn't use the same model for everything. That would be wasteful.
I've got one prompt that I can execute in Claude Code. That is my planning prompt, where I tell it I've got a whole set of requirements. You're going to ask the user questions to clarify the requirements. Then you're going to propose what the plan might look like to implement this thing.
The planning phase uses cutting-edge models like Gemini 3 Pro, Claude 4.5 Sonnet, Claude 4.5 Opus. These expensive models excel at the creative, architectural thinking required to design solid systems. They research libraries, identify gotchas, and critique approaches.
Once you've got a really solid plan, in a lot of cases, you can actually use one of the less expensive models to execute the plan. They'll follow the steps really well, but they won't have the size and the scale or the creativity to look at the bigger picture.
This isn't just about saving money. It's about matching model capabilities to task requirements. The big models see patterns and possibilities. The smaller models execute instructions reliably. Eden has built a library of prompts - one for planning, another for architectural review, others for security checks and code quality verification.
You're kind of just building it up over time. Take a first stab at it and see where the AI screws up, and start writing your prompts around that.
The workflow mirrors good software engineering practice: specification first, then architecture review, then test-driven development. The AI doesn't naturally follow this process, so the prompts force structure onto tools that would otherwise jump straight to coding.
Annual Reset Part 3: Lessons from a 50 Year Old Book About Tennis
A tennis coach in 1974 identified the performance gap that most engineering education ignores: the battle between your judgmental mind and your capable self. W. Timothy Gallwey's "The Inner Game of Tennis" introduces the Self 1 vs Self 2 framework, where Self 1 is the critical voice convincing you to stay quiet in meetings and hide failures, while Self 2 is trying to actually learn and solve problems. The shift from treating curiosity as a liability to viewing questions as collaborative problem-solving can accelerate career contribution more than any technical training. Gallwey's insight about non-judgmental observation applies as much to debugging sessions and design reviews as it does to tennis courts, and the distinction between appearing competent and becoming capable matters more than most engineers realize.
To read the full article, visit The Wave.
When Your Machine Shop Says No
Last year, a customer couldn't find a shop willing to make their part. Difficult material, tight tolerances, small quantity.
We tested our internal supply chain. Made the parts ourselves. Sent them as a surprise.
They worked perfectly. That customer has ordered hundreds more since.
Now available: custom machined parts with Pipeline engineering oversight, pricing that outperforms domestic shops, and the same supply chain we trust for our own work.
Want to learn more? Contact us at the link below: