I want to be specific about something upfront: I'm not going to tell you that GenAI "changed everything" in a vague, inspirational way. What I'm going to do is tell you exactly what I did, what the results were, and what I got wrong at first.
I started integrating ChatGPT into my workflows in 2022 — when I was founding datalitiks. Not as an experiment. Not because it was trendy. Because we were a lean team with an ambitious product roadmap, and I needed to find a way to produce more without burning out or hiring people we couldn't afford.
What I actually changed
The first thing I did was audit every recurring task I spent more than 30 minutes on in a week. Writing, summarising, formatting, explaining, structuring — anything that involved producing text or organising information. That list was longer than I expected.
Then I systematically tried using GenAI for each one. Not all at once — one workflow at a time, over several weeks. I kept a simple log: what I tried, what worked, what didn't, how much time it saved.
The ones that worked immediately: drafting first versions of any document (proposals, reports, emails, frameworks), summarising long inputs (research, meeting notes, articles), and structuring my thinking when I had a complex problem but hadn't yet organised my approach.
The shift that made the biggest difference was treating GenAI as a first draft machine, not a final answer machine. Every output needed my expertise to review, refine, and make genuinely good. But having a solid first draft to react to — rather than starting from a blank page — changed the pace of everything.
The results — honestly
These numbers are real. They're also not magic — they came from deliberate, systematic adoption, not occasional use. The difference between someone who uses GenAI occasionally and someone who has embedded it into their workflows is like the difference between someone who owns a gym membership and someone who actually goes three times a week.
What I got wrong at first
I spent too much time in the early weeks trying to get GenAI to produce perfect outputs in one shot. It doesn't work that way. The prompts that produce good results are specific, contextual, and iterative — not one-and-done.
I also underestimated how much domain expertise matters. GenAI in the hands of someone who knows what good looks like is a multiplier. GenAI in the hands of someone who can't evaluate the output is a liability. The tool amplifies whoever is using it.
The practical advice I'd give to anyone starting: pick one task you do every week that involves writing or structuring information. Try doing it with Claude or ChatGPT for four weeks. Keep track of the time. Then decide whether to expand.
Don't start with a strategy. Start with one workflow. The strategy will become obvious once you've experienced what's actually possible.