Envelope #45: Predicting the future (sort of)

Good morning!

Andy from Back of the Envelope here.

Today, I’m predicting the future.

Well… kind of.

With AI evolving at breakneck speed, do you ever get that weird mix of excitement and unease?

Those are the kind of questions I've been asking… and I've been putting a lot of thoughts into this.

And today, I want to share some of those thoughts, because the better we understand what’s coming, the better we can position ourselves ahead of the curve.

Let’s go!
(Estimated read time = 3m 20s)

1. The Relentless Progression of AI and the Rise of the “Pseudo” Engineer

Remember last week when I shared how NotebookLM helped draft a concrete repair recommendation? And then I joked that AI could also write a song about it?

Well… turns out, that wasn’t a joke.

I took those lyrics, fed them into the latest ChatGPT model (GPT-4.5) to enhance them, and then used Suno v4 (song-writing AI) to generate a fully produced J-pop-style song— with vocals, music, everything.

And now, I can’t get this concrete repair track out of our heads!
(link in the bottom — check it out after reading the article)

This got me thinking…

AI is progressing at a ridiculous pace.

First, it was just text generation. It got really good, and at times, almost mind-blowing.

Then there are images, coding, audio and video — each rapidly improving.

Now, the hot topics are AI agents (AI doing real tasks like deep research and compiling reports) and robotics (have you seen the NVIDIA’s “Blue” robot from yesterday!? Link below also).

What does this mean for us?

It seems that the barrier to entry for any profession is getting lower, at least at the conceptual or schematic level, if we use AEC language.

There are already some AI-assisted software for architecture and land development that can generate building layouts rapidly (links below).

It’s only a matter of time before a generative AI tool for structural engineers shows up.

Maybe it produces rough framing concepts.
Maybe it suggests column placements.
Maybe it lays out preliminary lateral systems.

The generated result may be conceptual and most likely inaccurate, but convincing enough to an amateur.

And that’s the shift we need to be aware of.

Imagine this:

An architect or developer runs an AI tool that spits out a structural concept before talking to an engineer. They show up with AI-generated structural plans, thinking they’ve already done some of the work, and ask:

"Hey, since we already have a concept and schematic design, can we skip those phases and get a lower fee and start with design development or construction document?"

Sounds far-fetched? Maybe.

But we should be thinking about it now, because AI-generated content is going mainstream.

Soon, we won’t just be designing… we’ll be verifying, guiding, and correcting.

And that changes the conversation.

The question isn’t "Will AI replace us?"

More like, "How will we differentiate ourselves from what AI can do?"

2. The K-Shaped Economy: Enhancing vs. Replacing

Got this idea from entrepreneur Shaan Puri (worth a read - link below).

Here’s the gist:

A lot of people think AI will replace jobs.

But that’s not quite right.

AI doesn’t replace jobs.
It replaces tasks, or job skills.

And that’s an important distinction.

Think of Our Job as a "Bundle of Tasks"

Instead of seeing our jobs as one big, fixed thing, imagine breaking them down into a collection of tasks.

Take the RFP phase, for example. We:

• Review the RFP.

• Identify structural concerns.

• Communicate clarifications.

• Define the scope.

• Estimate and develop fees.

Each of those can be broken down even further.

Some tasks, like summarizing an RFP or identifying key requirements, can already be enhanced with AI.

But most of the tasks we do?
AI can’t replace them — yet.

How This Connects to the "K-Shaped" Future

Right now, most structural engineering tasks fall into three categories:

1. Unaffected – AI can’t help or replace.

2. Enhanced – AI can assist but not fully replace.

3. Replaced – AI can automate.

Right now, most of what we do is in the "Unaffected" category.

But over time, as AI improves in computer vision (understanding what's on screen and in drawings) and agentic workflows (e.g., AI doing things on our computers for us), more tasks will shift from Unaffected to Enhanced, and eventually to Replaced.

Which brings us to this:

The Future Value Shift

In addition to continually keeping up our engineering knowledge and expertise, one of the highest-value skill in the next 5–10 years will be managing AI tools.

Because when AI starts taking on more and more tasks, the engineers who know how to integrate AI into their workflow will become more valuable than those who ignore it.

Put another way, embracing AI isn’t about giving in to technology.

It’s about using it to level up our expertise.

Ahh, there’s so much more to explore, like “how do we embrace AI”?… but this email is getting too long so I’ll save it for another one.

Hopefully this sparks some ideas and gets you thinking more about it (ideally with more excitement than worry 😆).

Until next time,

Andy Lin
Back of the Envelope/Structural Engineer HQ

P.S.

If you haven’t heard of our sponsor, Durafuse, I put together a quick, 30-second read on how this moment frame connection benefits us structural engineers.

Take a look and let me know your thoughts!

P.P.S.

Here are the links I mentioned if you want to check them out:

• Concrete Strong (Suno AI Song)

• NVIDIA's Blue Robot

• Hypar

• Testfit

• Shaan Puri’s K-Economy Article