"Hello?"
"Hey JJ, it's Jarvis. I know you're on your way to a lunch meeting with Bob. I just wanted to remind you that it's Bob's birthday today."
"Good catch, Jarvis! I totally blanked on the reservation. Can you see if you can snag a table for two?"
"I’m on it. I’ll text you the details in a sec. While I’m at it, should I start drafting that proposal due next week?"
"Yes. Be sure you do both a Word doc and a PowerPoint."
"No problem, I will email you a draft by 3 p.m."
“Thanks Jarvis. Is there anything else I'm forgetting?"
"Of course there is…"
A few weeks ago, I decided to see if I could build my own AI version of a personal assistant. I wanted to know if the hype was actually useful for someone like me.
I started with three simple rules for my AI
personal assistant:
- It had to be extremely secure. With access to my calendar, email and selected work files I wanted to sleep at night knowing my personal AI assistant came with a personal AI security guard.
- It had to be able to talk on the phone. I spend a lot of time behind the wheel, so talking is always better than typing.
- It had to actually know me. No generic robot answers, it needed to understand how I work.
Truth be told, I’ve never really considered myself a "personal assistant" kind of guy. I know plenty of successful people who rely on personal assistants and the setup works well for them. But for me? The whole idea always felt a bit too over the top.
Still, the hottest new technology has a funny way of nudging me until I try it. So, I dusted off my geek hat, ignored the "too fancy" alarm bells and decided to dive headfirst into the deep end.
The job description for an AI assistant isn’t that different from a human one. You want someone trustworthy, organized and proactive. They need to be ready to help when things get busy and look for ways to make your day run smoother.It took me a few weeks to get the first version running — I'll skip the boring tech parts, but using AI to build other AI is pretty wild).
Since then, I’ve been constantly tweaking it. I give it specific instructions. I show it how I like my notes formatted. I give it context on my current projects. And then, I just talk to it like a regular person:
“What’s the most important thing on my to-do list today?”
“Is there anything I’m forgetting?”
“Help me prep for this 2:00 meeting.”
“What did I say I was going to do next?” (That last one is a lifesaver.)
The more I use it, the better it gets. We’re learning how to communicate with each other. And just like managing a team, it all comes down to how you ask. If I say, "Help me get ready for this meeting," it might give me a five-page report I don't have time to read.
But if I say, "Give me a one-minute briefing with three key points and what I should ask," I get exactly what I need. It turns out clear communication works well whether you’re talking to a computer or a human.
It’s not perfect, of course. AI can still get things wrong, and there are limits to how much I trust it. Privacy and security have to be baked into every step, and I’m still careful about letting AI take action versus just recommending it.
For now, I still want a person making the final call. Luckily, I’m usually around for that.
Still, I can see where this is heading. We’ve spent decades working around our computers — digging through folders, searching for emails and hopping between apps just for one piece of info. My AI assistant flips that script. Instead of me hunting down data, it brings it to me. I’m still early in this experiment, but I’m hooked. Turns out, having a little extra help isn’t so “fancy” after all, it’s just a smarter way to work.
JJ Rosen is the founder of Atiba, a custom software development firm, and Nashville IT support company. Visit Atiba.com for more info.
This article originally appeared on Nashville Tennessean: I built an AI personal assistant. Now I'm hooked.
