There is always another list.

There is always another list.

“10 books every software engineer should read.”

“15 books every architect must know.”

“25 books that will make you a better developer.”

I understand the intent. There is a lot of wisdom in those books.

But I am not reading 15 books.

I am probably not reading five.

And I do not think that is the best learning strategy for many experienced engineers anymore.

I have been in software engineering for decades. I have taken the classes, built the systems, worked through the failures, led teams, changed technologies, and read hundreds of books over the years.

At this point, the problem is not lack of information.

The problem is finding the right information at the right time.

That is where AI has fundamentally changed professional learning.

We Used to Learn Before We Built

The traditional model looked something like this:

Read the book.

Take the class.

Learn the framework.

Study the architecture.

Then eventually apply it.

That made sense when knowledge moved more slowly and information was harder to access.

Today, I can start with a real engineering problem and ask AI to pull together the most relevant ideas from decades of software engineering knowledge.

That does not mean blindly accepting an AI summary.

Experience still matters.

Judgment matters even more.

But the learning loop is different.

Start With the Problem

Suppose I want to improve how engineering teams deliver software.

I do not necessarily need to spend the next six months reading five major engineering books from cover to cover.

I can start with five excellent sources:

• Accelerate

• Team Topologies

• Designing Machine Learning Systems

• Building LLM Powered Applications

• The Pragmatic Programmer

Then I can ask a much more useful question:

What is the 20% from each of these books that still matters today?

What has AI made less important?

What principles have become even more important?

And what can I actually apply Monday morning?

That changes everything.

The Durable Lessons Are Surprisingly Simple

When you compress decades of engineering thinking, many of the strongest ideas survive.

Measure outcomes, not activity.

Reduce cognitive load.

Design systems for change.

Automate repetitive work.

Keep feedback loops short.

Make responsibilities clear.

Build reusable capabilities.

Test assumptions.

Observe production systems.

Separate reasoning from verification.

Those principles are not obsolete.

In many cases, AI makes them more important.

AI Changes What We Need to Memorize

Software engineers spent years learning syntax, APIs, frameworks, design patterns, configuration details, and implementation techniques.

Those skills still have value.

But their value is changing.

AI can increasingly generate boilerplate, explain unfamiliar code, perform refactoring, search documentation, create tests, explore repositories, and implement routine features.

The scarce capability is becoming something else:

Knowing what should be built.

Knowing what constraints matter.

Knowing what evidence proves success.

Knowing when the system is wrong.

Knowing what should be reusable.

Knowing what should be automated.

Knowing what should remain under human control.

That is not less engineering.

It may actually be more engineering.

Experience Becomes More Valuable, Not Less

This is where experienced engineers have an enormous opportunity.

Thirty years of experience cannot be compressed into a prompt.

Experience gives you pattern recognition.

You have seen systems fail.

You have watched teams over-engineer simple problems.

You have seen technologies disappear.

You know that the newest framework is rarely the real answer.

AI can give you enormous breadth.

Experience helps you determine what deserves your attention.

That combination is powerful.

Build in Order to Learn

The learning model I increasingly prefer looks like this:

Problem → Intent → Research → Build → Test → Evidence → Reflection

Start with a real problem.

State the outcome you want.

Use AI to research the strongest existing ideas.

Build something.

Test it.

Capture evidence.

Then reflect on what worked.

If an important topic appears along the way, go deeper.

Read the chapter.

Read the paper.

Read the book if necessary.

But now you are reading with purpose.

That is very different from consuming thousands of pages because someone put them on a graphic.

Books Still Matter

I am not arguing that books are obsolete.

Far from it.

Books provide depth, reasoning, context, and perspective that short-form content cannot reproduce.

Sometimes a great book changes how you think permanently.

But I no longer believe the default advice to engineers should be:

“Here are 15 books. Start reading.”

I think the better advice is:

Build something difficult.

Use AI to access the best thinking available.

Go deeper when the problem demands it.

And retain the principles that survive contact with reality.

The New Advantage

The engineer who wins in the next decade may not be the person who has consumed the most information.

It may be the person who can turn information into useful action the fastest.

That means combining:

experience,

judgment,

AI,

experimentation,

and evidence.

The library still matters.

But now we have something new.

We have an intelligent research partner sitting beside us.

Use it.

Do not stop learning.

Just stop assuming that learning has to happen before the work begins.

Build in order to learn.

I’d probably use “