EG170: stop asking if AI will kill coding

Software development is too complex. 

To be a good software engineer requires a lot of experience and practice. 

AI will never be capable of fully replacing developers.

I have heard this argument in many different contexts and teams in the past 2 years.

While I don’t have the technical expertise to debate, agree or contradict this line of thinking, I am now convinced that’s a false discussion.

AI doesn’t need to fully replace the skills and expertise of individual, human software engineers to have a significant impact on software development.

It’s enough if it changes the process and the business models.

Between 2017 and 2020 I have visited more than 80 factories in Europe, US, Mexico and China.

I was selling complex industrial automation technology for automotive and furniture manufacturing. The full solution had hardware, software, customizations, integrations, deployment, training and support services.

I have also discussed during that period with representatives from at least another 320 companies (process engineers, manufacturing managers, owners and so on).

I have seen every possible level of automation you can think of.

From the most advanced automotive factories, using robots and fully integrated production lines.

To the most primitive, with hundreds of manual workers, in cold, dark factories, cutting leather for sofas with scissors.

And everything in between.

In that world, I have also heard people saying that software and algorithms cannot replace skilled workers, those with 10 or 20 years of experience.

It was true. In some very specific contexts, for particular types of products and manufacturing processes, the skilled manual labourers were better at their tasks when compared 1-to-1 with software algorithms.

This was probably 1 to 2% of the hundreds of cases I have seen.

In all other cases, automation was significantly better.

And it was always better for full production lines, where all the steps were digitized and automated as much as it was technically and economically possible.

This is what I am already seeing happening with AI and software development.

Here are 2 examples, from Daniel Ilinca on product managers using AI to create POCs and Dan Boresjo on AI hijacking tech stacks.

The key is not in individual skills.

Software engineers are not paid based on their skills.

The single most important variable that determines how much software engineers are paid is their employer’s business model.

It’s all about the model, the system, not the individual projects or tasks.

And that’s changing. Fast.

> AI is enabling more modular and automated code generation, similar to how assembly lines broke down complex tasks

> AI-powered tools are improving code review, testing, and deployment, leading to faster development cycles

> AI is facilitating the creation of low-code/no-code platforms, allowing non-developers to contribute to software development

On December 1, 1913, Henry Ford installed the first moving assembly line for the mass production of an entire automobile. His innovation reduced the time it took to build a car from more than 12 hours to one hour and 33 minutes. Ford broke the Model T’s assembly into 84 discrete steps and trained each of his workers to do just one.

History doesn’t repeat itself, but it rhymes.

============

PS. 

On Saturday, March 1st, I spent 5 hours on Zoom with the 2 co-founders of a team from Eastern Europe that has been doing MVP development for startups.

It all started on Thursday the week before, when they saw me in a webinar called “Segment and conquer”.

The day after the webinar they sent me an email with the subject line: “Your presentation was eye-opening”

We had a short call the next Tuesday and I sent them an offer.

1 hour later they emailed back: “Let's do the workshop. Please send us the requirements and what details you need from us along with the invoice and the earliest day possible”

On Saturday we spent 5 hours together on Zoom.

Below is what they signed-up for.

Message me if you want to do this strategy workshop with your team.

Strategy workshop: choosing the right customers segments for your team

Format:

4h workshop (online on Zoom), maximum 8 participants

Input from you:

↳ Participants answer a short survey before the workshop

↳ Send any relevant info with your current commercial context

Content:

Theory, practical concepts and case studies based on your questions and challenges.

Targeted growth is critical for a thriving software agency.

↳ The specific client profiles you prioritize

↳ The messaging that resonates with those segments

↳ How your team's expertise aligns with their needs

↳ How you adapt your sales and marketing strategies

All these aspects determine your success and should be defined strategically.