Andrei KalamkarovI lead an AI-native team behind Concordia University's web and mobile platforms, and write about what daily AI use actually looks like inside higher ed, where change is real but rarely fast.
- 10 Jul
The memory I'm building for AI
AI forgets you between sessions. So I build its memory by hand: plain-text files that hold the decisions, workflows, and traps only I know. Here's the debate about how much to write down, what the early evidence says, and the files that earn their place in my work.
- 05 Jun
Knowing when to stop
When AI took the work out of iterating, the friction that used to stop me vanished. The surprise wasn't the wasted time. It was where the cost actually went.
- 28 May
Sensitive data doesn't mean no AI
Most people think sensitive data and AI don't mix. The real answer is different habits, different tools, and different boundaries. Here's what that looks like in practice.
- 11 May
Side quest: bring an old project back to life
One good way to start with AI is to revive something old: a dead website, a broken archive, an old blog, or a lost portfolio.
- 23 Apr
Teaching AI to sound like me
AI revision tends to make your writing sound like everyone else's. A simple profile file keeps the voice yours, and ends up improving it too.
- 17 Mar
AI doesn't hallucinate your photos. It edits them.
I fed the same photos through ChatGPT, Copilot, and Gemini. Each one quietly decided what to keep, what to erase, and what to invent.
- 24 Feb
Signal Drift: what happens when four AIs build one app
Can you get better answers by switching AI models instead of sticking with one? I built an app to find out.