Ivan Kahl

🇿🇦 South African software engineer building modern software with AI. .NET, AI engineering, automation, and developer education.

[John Crickett @johncrickett]

My prediction for software engineering over the next 3 to 5 years.

Computer science and software engineering fundamentals will become far more valuable than knowing the latest frameworks.

If AI can write code in almost any language, knowing the nuance of another framework becomes less of a competitive advantage.

But understanding algorithms, data structures, networking, operating systems, databases, architecture and security?

That's a different story.

You still need to know:

- Why a particular architecture is appropriate.
- Why an API might fail.
- Why a database won't scale.
- Why an apparently working solution might be insecure.

A coding agent can build a solution.

But you need the knowledge to evaluate the trade-offs and recognise when it's wrong.

So, I think we'll see a renewed emphasis on the fundamentals of computer science and software engineering.

TL/DR: Learning how computers and software systems work will matter more than memorising how to use them.

What do you think?

[John Crickett @johncrickett]

“AI will replace software engineers because it can write all the code.”

Nope.

Several years ago I built software that unlocked more than $10M a year in additional profit for the business.

The code took less than two weeks.

With today’s AI tools, I could probably do it in under a week. Maybe even a day.

But the code wasn’t the hard part.

The hard part took months:

Understanding a messy domain.

Pressure-testing assumptions with experts.

Working through edge cases.

Deciding what not to build.

Making tradeoffs around risk, complexity, and value, then getting stakeholders to sign off on them.

By the time I wrote most of the code, much of the real engineering had already happened.

That’s why “AI can write code, therefore AI can replace engineers” misses the point.

Code is an output of engineering.

AI makes that output faster and cheaper.

But someone still has to understand the problem, challenge the assumptions, make the tradeoffs, and decide what the software should actually do.

That’s the engineering.

[Uncle Bob Martin @unclebobmartin]

True. This does not imply that you have to review every line. Lots of systems have had full code reviews and still should not have been deployed.

Test, test, test. Measure crap. Do mutation tests. Check dependencies. Look at the module structure. Use the agents to interrogate the system. Verify, verify, verify.

The productivity win is big; but the risk is high unless you are very diligent in using the best tools at your disposal; efficiently enough to keep that win.

[David Fowler @davidfowl]

I’ve been running a few 24/7 agents on code bases for the last week or so on various codebases and I can say with certainty that’s there’s so much stuff that does not meet the priority bar of most Eng teams that fly beneath the radar until it blows up in some more meaningful way.

[Ivan Kahl @IvanKahl]

@toddmotto I've found the opposite. With AI I can actually solve problems far better without being slowed down by writing the actual lines of code.

[dax @thdxr]

solve your own problems. it's so much easier to build a good product you use every day. i only build products i need

just make sure it's something a lot of people can one day use

[Thariq @trq212]

very weird thing happening rn

technical people are sad, they think their skills are now useless

non-technical people are desperately wishing they were more technical so they could use agents better

[Ivan Kahl @IvanKahl]

@CFDevelop A lot of people in tech aren’t at the cutting edge of AI yet. When someone makes that claim, I assume they’ve just started with agentic coding. The mistake is letting first impressions stop them from learning to use it well.

[Thariq @trq212]

@hnshah I think all technical people should be massively amplified, there are so many ambitious problems to solve. But it is a big change in mindset.

[John Crickett @johncrickett]

My prediction for software engineering over the next 3 to 5 years.

AI won't replace software engineers. It'll drive even greater demand for them.

A team of 5 experienced engineers using AI will outperform a team of 20 pre-AI.

Which will dramatically reduce the cost of building software.

I think we'll see Jevons paradox in action. As something becomes cheaper and more efficient to produce, demand for it can increase dramatically.

Thousands of software projects that previously weren't economically viable will become worth doing.

The winners will be companies, big and small, that behave like Google, Amazon and 3M, with a culture of experimentation, a willingness to try lots of ideas and the discipline to kill their darlings.

TL/DR: Smaller engineering teams, more innovation, more software products, and potentially more software engineers than ever.

What do you think?

[Theo - t3.gg @theo]

https://t.co/cRdL7orwcq

[Rhys @RhysSullivan]

the models love to add so much extra code for cases that will never happen

one example is in review it 'caught':

"changing a setting while being demoted from admin in the same request" which no user would ever hit

but then they miss the obvious things that grind down your codebase quality over time

[Gergely Orosz @GergelyOrosz]

Going thru all responses. Such wild variance!!

General trends:

- PRs per eng is up
- Spend is anywwhere from $100/dev/month to ~$5,000. A few outliers in Big Tech at $20K+
- Very very very few teams ask about the ROI on the spend!! Or know how to measure. More want to tho

[Gergely Orosz @GergelyOrosz]

I don't think most people realize how complex and complicated agentic AI systems inside mid-sized and above companies are: by necessity, because there is SO MUCH context, but only so little AI contex windows.

Case study on Uber's 19 internal, custom, AI-related systems coming later today in @Pragmatic_Eng

So many of those "AI-related systems" are "systems pre-AI" that turned out to become extremely useful with AI. Like their PRD+ERD system that conveniently documents connections between the ~5,000ish internal systems inside the company!

[Ivan Kahl @IvanKahl]

@GergelyOrosz Curious about this as well. But rather than PR output, you really should be measuring DORA metrics to get a proper idea of how much value it’s actually producing.

[Ivan Kahl @IvanKahl]

Slop existed long before generative AI. Unfortunately, AI removes a lot of the judgment and review that previously took place when creating content or writing code. That’s not AI’s fault; it’s the human’s fault.

https://t.co/UVTq7oKE4z

[Aaron Stannard @Aaronontheweb]

People on X absolutely have their heads so far up their own asses about AI use in the real world.

Most BigCo tech employees are on CoPilot auto-complete prompts or copy-and-pasting shit from ChatGPT's web UI, and even that is unauthorized by the BigCo Boomer Council of IT Elders

Many BigCo apps still don't even have automated testing, CI/CD, or containerization, so no building blocks AT ALL for even starting with agentic coding.

And these aren't dipshitty little micro-SaaS apps; they're 9-figure businesses that are often invisible to your day-to-day existence as an anxious shit-poster on this website.

The AI revolution around coding is still very much in its early adopter phase. The "normie knowledge worker" AI adoption process hasn't even really started yet

[Ivan Kahl @IvanKahl]

@davidfowl These are awesome!

[Ivan Kahl @IvanKahl]

@codeopinion Agreed, and this is one area where I think using AI agents can be problematic. They will always build something and, if you aren't intentional about design early enough, it'll make the decision and choose the trade-offs for you... and they're often not the best ones.

[Ivan Kahl @IvanKahl]

@LukeParkerDev haha this is awesome! love it!