Solve First, Then Ask AI: How a Year of Daily Coding Challenges Taught Me More by Delaying the Help

freeCodeCamp Community
,
1

Hi everyone,

Over the past year, I worked through freeCodeCamp’s Daily Coding Challenges one at a time, every day, focusing on Python solutions, and kept my work in a public repo. Along the way I landed on a workflow that made a real difference in how much I actually learned, and I wanted to share it here in case it helps other people working through the DCCs.

The rule I set for myself was simple: solve the challenge completely on my own first, with no AI assistance at all. Only after I had a working solution did I bring in AI — and even then, not to write code for me, but to get feedback on what I’d already done, see alternative approaches I hadn’t considered, and dig deeper into concepts the challenge had touched on.

That ordering turned out to matter a lot. Struggling through a problem on my own — dead ends included — is what actually built the skill. Using AI afterward, once I already had skin in the game, turned it into a genuinely useful tutor: it could point out where my approach was clunky, show me idioms I didn’t know, or explain why a different solution worked better, instead of just handing me the answer before I’d done the thinking.

One side effect I didn’t fully expect: sticking with Python every day, rather than also doing the JavaScript equivalents, pushed me to actually study the standard library instead of reaching for whatever felt familiar. A lot of the “alternative approaches” I discovered afterward turned out to be built-in tools I hadn’t known existed.

If you’re working through the daily challenges and wondering how to use AI without shortcutting your own growth, I’d encourage trying this: close the AI tab until you have a solution, then reopen it to review and extend what you built.

I wrote up the full workflow in more detail in my repo’s README, if you want to see how I structured it day to day: GitHub - karelplanken/fcc-coding-challenges · GitHub

Happy to answer questions if anyone wants to try something similar.

6 Likes
2

Congratulations on completing all these! And thanks for sharing your experience here and what you learned along the way. I’m CC’ing @moT01 who designed the challenges so he can see this, too.

2 Likes
3

Congratulations @karelplanken :tada: That’s quite the achievement - and definitely a good use of AI to help you learn :clap:

Just out of curiosity, do you recall any specific challenges that you liked a little extra? I’m quite proud of many of them, but “Left-handed seat at the table” was particularly fun to create for me.

Also, what are you going to do now that there aren’t new challenges every day?

2 Likes
4

Dear moT01, first and foremost thank you for designing these challenges and thank you for reaching out. I haven’t noted down specific challenges that I like better than others, but I really like the challenges that can be solved in just a couple of lines particularly when using the “batteries” that come with Python (standard libs). I did make a list of the harder challenges and those usually involve general CS methods (DSA), like for example backtracking and so on. I must say, however, when looking at my list a number of challenges has lost their toughness to me . Apparently my skills have improved :smiley:…

My plan for the upcoming year is to review and refactor my solutions to the challenges, straighten things out so that they meet all the strictness I impose on them (and myself of course :wink: : please see pyproject.toml, additionally I have type checking mode set to strict), Recently I released a couple of tools on my scripts I orchestrate with just, inspired by (and started from) Nacho Lorca’s mold who had a talk on Europython 2026 in Krakow Poland I attended.