In this AI era, coding has become exponentially accessible. Yet, it doesn't mean you can do everything you have in mind with AI.

Computational design is at the intersection of engineering and tech (by engineering I naturally mean any AEC professional). So, there are two technical knowledge and skills you should have or develop:

• Coding skills & knowledge

• Engineering skills & knowledge

I will assume you're an engineer so we can focus on the tech part. Software engineers on the other hand, need to learn how engineering works, and this my friends take a huge amount of time.

So I prefer to bet on engineers who are willing to code!

Find a mentor/community

With so many available trainings and tools, it can be time-consuming to find what you truly need to use. Worse, you may already be learning some Grasshopper, but not learning components you may use for your own discipline.

Having a mentor or being part of a community dedicated to computational design can help orient your learning and accelerate your journey significantly.

Prioritize people in your own discipline, they will be able to speak your language. Good thing, PatchWork has been working hard to gather computational designers and aspiring computational designers from different disciplines. Join us to find a mentor and obtain clear guidance.

Vibe code a very simple script for you

Your mentor suggested you to start with Python in Sofistik? Great let's start with that. With AI, it is incredibly easy to start from scratch. Identify a very simple task and code it with AI! Setup everything with Claude if necessary and prompt directly what you want to achieve. You get your first code, hopefully running properly.

Analyze and acquire the technical knowledge

With your first code in hand, the real learning starts now. You didn’t understand much of what Claude was typing? It's time to take a break in the rush and strengthen your fundamentals. The whole purpose of rushing with AI was to have a situation where the code outcomes would make sense to you and your daily duty.

1. Algorithmic thinking

Always with your bestie, prompt it to explain step by step what it did and how it wrote it. What we want now is to understand the logic of the code. In other words: the algorithmic thinking behind what's written. Try those prompts:

"Detail the written code step by step. Explain what is the goal of each step and how you did it, describe all necessary functions or classes needed to understand the steps." This, my friend, is your first entrance into algorithmic thinking: Dividing a goal into smaller goals that can be achieved by clearly defined step. Because a machine doesn’t think, it repeats actions. By doing so, you get closer to creating a solution that works, rather than something that seems to work.

2. Syntax, Types, Functions, etc…

The previous step gave you the plan, the actions that need to be done. Now, we need to understand how to make them happen and this is where you dive in the pure coding aspect: learning the syntax of a coding language.

Whether it is Grasshopper, Python or C#, the logic is the same.

Creating a cross-section for example requires to create points that will form a polyline.

• In Grasshopper, you would have the component Construct Point

• In Python, you may create a function def createPoint(x, y, z)

• In C#, you may have a Point3D class which props are int x, int y and int z

Read it as many times as needed until you understand 100% what is happening with these lines of code. Then, type the exact same code. Just to get used to it.

Recreate the same solution, by hand this time

Next steps is to consolidate what you learnt so far. Recreate the exact same solution but without any AI help. It doesn’t matter if it takes you 5 minutes or 3 days to recreate it. What matters is that you manage to make it work yourself. It will consolidate your knowledge and push your brain to adopt machine thinking.

Refacto

In computational design, there is not a unique solution to solve a problem. There is an infinity of ways. They all have their own pros and cons and an experienced computational designer is sensitive to this. Refacto means to redo all your code but in a "better" way. Logic will change but the goal and outcome should be nearly the same (or better!). 

This time, I advise to not start right away with AI. Take a pen and a notebook, write the different steps you can imagine and then try to type it. You may use AI at that moment to help you with the syntax.

What do I want to develop in you by refactoring? Creativity. A good computational designer has a constant flow of creativity while coding and honestly, sometimes I spend more time writing down all the solutions I have in mind than actually bringing them to life.

Upgrade

Finally, you have laid the foundation of your solution, your first feature is working extremely well, and now it is time to expand your core features. Select another useful feature and make it happen! You already know what I am gonna say: repeat what you learnt so far. The process is exactly the same.

So this is how I would personally learn computational design in 2026 as an engineer. AI truly lowered the barriers to enter computational design. Some might stick to vibe coding, but a vibe coder will never match someone who sweats to understand coding fundamentals.