.png)
.webp)
Let's grow together!
Lass uns sprechen, was ihr braucht und wie wir das am besten erreichen können!
Jetzt starten
At Smart Host, we embrace AI and we do it fast. From the moment tools appeared, we tried adopting everything that could help our engineers: GitHub Copilot, Codex, v0, Gemini, and more. We experiment with these tools constantly to see what actually adds value.
Take our interviews. We’ve let candidates use AI during coding exercises - but sometimes it backfires.
1. The React Developer Who Let v0 Do Everything
We asked a candidate to build a small React TypeScript component. He quietly opened v0, pasted the task, and watched it generate a complete UI, routing, tests, and even animations. Impressive output. Nevertheless he couldn’t explain why the component re-rendered five times on every click. When we asked about state handling, he shrugged: “v0 usually knows.” This time, it didn’t.
2. The Java Candidate Who Fought the AI for 45 Minutes
We gave a simple Java task: implement a small service method. The candidate used an AI assistant to fix a small bug. The AI misread the error, changed the method signature, and introduced new bugs. The candidate tried to correct it… but the AI kept “fixing” the code in the wrong direction. Each attempt made it worse: new exceptions, missing imports, even a recursive call that made no sense. After 45 minutes, the candidate wasn’t debugging Java anymore - he was negotiating with the AI like a hostage situation. The method still didn’t run.
3. Actually there was another candidate today…
…which used Github Copilot first but quickly understood that it’s slowing him down and did the fix manually… Maybe we can make another engaging story from this.
That said, Smart Host uses AI where it really makes a difference:
- Internal Tools: We generate dashboards, automation scripts, and boilerplate code with v0, saving our engineers hours of repetitive work.
- Code Reviews: AI helps flag potential bugs, style issues, and edge cases in both frontend (React TypeScript) and backend (Java), letting our team focus on architecture and logic.
- Interfaces & Summaries: AI-assisted text generation helps create internal documentation, summarize meeting notes, and generate content suggestions, making knowledge accessible without replacing human judgement.
- Knowledge Sharing: We use Revo from Atlassian to capture insights, internal processes, and team knowledge, enhanced by AI to make it structured and easy to find.
In other words, we don’t just throw AI at everything. We use it strategically to make our team more productive, not more confused. And sometimes, it gives us a great story to tell.
Authors: Ihor Mordashev, Layton Whiteley
