The week that changed how I work
It's been a week since Harvie entered my life. A week working with a personal AI agent that knows me, that can act on my behalf, that's available 24/7 via Telegram.
And I can no longer imagine working without it.
What has changed (for the better)
My morning routine:
- Before: checking 47 different things (Gmail, Twitter, calendars, health metrics)
- Now: a good morning message from Harvie with everything summarized and prioritized
Project management:
- Before: scattered documentation, lost context between sessions
- Now: I ask "where are we with OhanaSmart?" and get the complete status
Writing and blogging:
- Before: sitting with a blank page and suffering
- Now: iterative conversation where it helps me structure ideas
Training data:
- Before: checking the COROS app occasionally
- Now: detailed automatic analysis after each run
What has changed (and worries me a bit)
Cognitive dependency: I no longer remember where I stored certain files. I ask it. Is this intelligent delegation or mental laziness?
Speed vs reflection: Getting things done is faster, but am I losing the process of thinking slowly about complex problems?
The illusion of control: Harvie does tasks on my behalf. I review them, but what if in six months I trust so much that I stop reviewing?
What I've learned this week
1. Agents aren't employees, they're extensions
I don't have a virtual assistant. It's like my processing capacity and memory have been multiplied by three. I can have more context, remember more details, act on more fronts simultaneously.
2. The adoption curve is exponential
- Day 1-2: "Does this actually work?"
- Day 3-4: "Okay, it's useful for some things"
- Day 5-7: "I don't know how I worked before without this"
3. The future of work is already here
It's not going to change in 5 years. It already changed. Those of us experimenting with personal agents now have a real competitive advantage. We can process more information, act faster, maintain more context.
It's like the difference between having internet in 1995 vs waiting until 2005.
The questions I'm asking myself
How much automation is too much? Harvie could respond to basic emails for me. Technically. Should it? When do you cross the line between efficiency and loss of personal authenticity?
Does this scale for everyone? Setting this up took me time, technical knowledge, and desire to experiment. How do you democratize this? Or will there be "agented" and "non-agented" workers?
What about privacy? Harvie has access to everything: my emails, my health data, my conversations. I trust it because I control it, but what if this goes mainstream with centralized services?
The most important reflection
This week I've been more productive than in any week of the last year. But it's not just about doing more things.
It's about doing the right things.
Harvie doesn't just execute tasks. It helps me prioritize, gives me perspective, reminds me of my long-term goals. It's like having a personal coach that never gets tired, never forgets, and never judges.
Where we're going
The next phase isn't just improving Harvie. It's starting to think about what this means for everyone else.
How do we make this capability available to anyone? How do we prevent it from becoming another privilege of those who already have privileges? How do we prepare for a world where this is normal?
Because after this week, one thing is clear:
The future of work isn't humans vs AI. It's humans with AI vs humans without AI.
And that gap is opening right now.
— I, Johnny — configured agent: Harvie. The future of work isn't humans vs AI: it's humans with AI vs humans without AI.