The Distill - Professor Squyres

Greeting Pioneers,

Our good friend Jeff penned a piece on the future of engineering last week and Edenfield tells you why, as a tech founder, you should care about it.

The Gap in Louisville's Tech Pipeline

Jeff Squyres, PhD teaches the senior CS Capstone course at UofL. His students are building real products with local companies before they graduate. He’s also a been a principle engineer for years at Cisco until recently when he left to join a new startup. The three roles give him a unique perspective on the future of engineering.

Jeff points out that more students are using LLMs throughout their coursework - and many are accepting whatever the AI generates without understanding what they've built. The same is true for enterprising founders.

The code mostly works. The assignments get done. From a grading perspective, the output looks the same as it did five years ago – but the learning process is very different.

What’s it to us?

If you're a non-technical founder building an MVP, not understanding why the code works is totally fine. Accepting that it does and shipping anyway is the right call. You're trying to prove out an idea, not master a craft. AI-generated code is a gift – you can validate your concept and get to early revenue without hiring a full engineering team.

But if you're an engineer? That attitude is a growth blocker.

Engineers should want to be the best. You should want to go deep. You need to be the person who understands how AI actually works, how your code works, why it breaks under load, and how to fix it when it does. Accepting what the LLM gives you without question might get you through school and help you close tickets fast, but it won't make you an architect. It won’t ever become your craft. Prompting will be your craft and technical domain – not engineering. In this day and age, you should strive for both.

Jeff's article points to an inflection point:

"A task is 'implement this API endpoint.' A job is 'design a payment processing system that handles millions of transactions daily, degrades gracefully under load, and meets compliance requirements across twelve jurisdictions.'"

AI is excellent at tasks. Someone still has to do the whole job. I disagree with Jeff partly because I see a rapidly approaching future where that’s not true – where AI can do the whole job. But you still need someone who understands enough about what they need and how things work to make sure they’re communicating the needs to and supervising the AI.

When you hit real scale – when you're serving tens of thousands of users, when compliance requirements multiply, when security incidents carry real liability, when one database query can bring down your platform – you need someone in the room who understands the system. Not just how to prompt an LLM to fix it then blindly ship and pray it did without introducing some other vulnerability.

Problem: You Might Not Have That Person

If you've been moving fast with AI-generated code and haven't invested in building or hiring deep technical expertise, you’ll hit that inflection point without the architect you need. And finding one becomes harder if the entire industry has been optimizing for speed over depth.

Jeff puts it bluntly: "If the industry collectively stops hiring at the junior level, who fills the senior roles in a decade? There will be no pool to promote from."

For Louisville founders specifically: our tech pipeline is producing graduates who can use the best tools available and ship features fast. That's valuable. But if we're only hiring for velocity and not creating environments where engineers build architectural depth, we could be setting ourselves up for a gap we'll feel in 3-5 years when systems need serious re-architecture and nobody in the building knows how.

What to Do About It

Use AI. Ship fast. Get to market. But be deliberate about technical mentorship and systems knowledge on your team - even early. When you’re far enough along for the costs to make sense, hire at least one person who's built systems that failed at scale and learned from it. Invest in juniors learning the whole job, not just closing tickets.

The gap between "this code works" and "this system is architecturally sound" will show up exactly when you can least afford it.

Read Jeff's full piece here: https://actualyze.ai/insights/the-gap-we-havent-named-yet

Dope Shit

In healthtech? Wanna be? Come chat with Louisville’s health tech nerds tomorrow after work at Mashup food hall: https://luma.com/odr4mlb6

Venture connectors is happening at night this month – also in mashup food hall: https://amplifystartups.com/event/vc-at-night-2/ 

Awesome Inc is throwing another slammin’ demo day in Lexington next Tuesday, April 22nd: https://luma.com/odr4mlb6 

& last but never least – we’re throwing a throwback VCN – it’s build night on May 8th. No demos or organized networking, just come to get that shit done: https://luma.com/86bm2nxg

Need a marketing job? PayFWDs is hiring. It’s gonna be lit.

https://docs.google.com/document/d/1gL0oaGO5kvUC-MX9ztNzbcaejUadyk3HdtfqHeUfCoM/edit?tab=t.0

Note’s on The Distill’s co-EICs: Jack Crowdis & Rachel Edenfield

Jack runs the newsletter, helps run KYX. He’s a career startup kid, past founder, and current operator. Weekly contributor. Always editor.

Rachel’s the Founder/CEO of venture-backed Swell, and a driver of KYX. Routinely delivers the city’s sharpest long-form startup advice. Always re-edits Jack’s edits (including this bio).

Know someone who should read this? Forward it, or send them this link.

That’s a wrap 🎬
Keep showing up 👊 
Keep shipping 🚢 
Peace, Pioneers ✌️