Day 100 of the 100-Day Agentic Engineer Challenge: What Changed?
Yesterday was the last day of my 100 Days Agentic Engineer Challenge.
Let’s go back to my first post from January 1st and check what it was all about:
Looking back, I can see my thinking was quite chaotic. I wrote that my goal was to become a developer who knows how to build a true AI Agent. But I also mentioned toolsets, AI-assisted coding, learning Python, Machine Learning — and even things like improving my sleep, health, and mental condition.
Too many goals. Lots of chaos.
But also — consistency.
I wrote a post almost every day. Sometimes I skipped a day, but I made up for it the next. I also kept exercising, which was one of my biggest challenges — my poor physical condition.
Another Person After 100 days
When I read that first post, it feels like it was written by a different person. Someone who was searching for something but wasn’t sure what path to take — yet deeply wanted change.
My biggest progress was definitely physical.
On Day 1, I couldn’t do even one push-up. My arms were too weak to hold my body weight.
Yesterday, I did 100 push-ups (in 2 sets of 50), then 100 squats, and 100 crunches — all in single sets. My original goal was 100 push-ups in a single set, but I’m happy with this result. I’ll keep training until I get there.
A Journey Through AI, Coding, and Tooling
This journey was full of tech: AI, coding agents, and building tools.
I didn’t fully build a complete AI Agent, but I did create components — like long-term memory.
My biggest tech milestone:
I moved from no-code to code, using AI tools like AI builders and AI-powered editors. I explored many paths, and now I’m developing my own ideas with AI-first tools like Cursor and intelligent IDEs. I apply the same approach in my agency — using AI to deliver better, faster projects for clients.
I’m still learning to code with AI support, but now I finally have a direction. I haven’t settled on the perfect tech stack yet, but I’m getting closer.
Here’s what I’m exploring now:
- Frontend: Next.js, Astro with React, Tailwind, Shadcn
- Backend: Supabase (thinking about self-hosted on DigitalOcean), FastAPI
- Other tools: PydanticAI, Logfire, Letta (Memory Management), LangGraph’s LangMem, and of course, n8n, which is probably the only low-code tool I still use often.
What’s Next?
My goal now is to find one tech stack that lets me quickly build and ship different MVP SaaS ideas.
For now, I’m thinking:
- Stack 1: Next.js + Supabase Cloud + Vercel hosting
- Stack 2: FastAPI + HTMX — super simple, clean setup
I know I sound a little chaotic again — like the person I was 100 days ago — but this time, I have clearer specs 😄
But honestly, tech stacks don’t matter that much.
What really matters: SHIPPING.
Products that work. Ideas made real. Value delivered.
So here’s the plan: I’m not starting a new “challenge,” but I’ll keep writing — every time I build something.
When I ship a product, I’ll write about:
- what I built
- what tools I used
- what I learned
- and most importantly — how it can help you, too.
Upcoming App Ideas
- n8n Article Writer: fully automated research & publishing to WordPress
- Voice Assistant: with long-term memory + time filtering (day, hour, month, range)
- Diffusion Model Fine-Tuner: train and generate (Flux)
- AI Shorts Generator: from story to YouTube
- Autonomous Online Magazine: with article & podcast generation
- Social Media Agent (X/Threads): that learns and posts on its own
Thanks for following the journey. Let’s keep building. 🚀
