Somewhere around 11 PM I found an obituary that filled in a gap I didn’t even know existed in my understanding of a family member’s history. That’s the kind of thing that happens when you let an AI loose on your contact database for a few hours.
I’ll be open: I’m not great at keeping up with people. Not because I don’t care, I do, but because life moves fast and the mental overhead of remembering who said what, whose birthday is coming up, what a friend is going through right now adds up in ways I don’t always manage well. I forget. I mean to reach out and don’t. I show up to conversations without context.
I needed a better system.
The foundation is Monica CRM¹ (Customer Relationship Management). Open source, self-hosted on my home server, completely private. It’s personal relationship management software, which sounds cold when you say it out loud, but the point isn’t to turn friendships into a sales pipeline. The point is to have a place where the context lives so my brain doesn’t have to hold all of it.
Klaus, my AI, spent an evening helping me actually populate it. We started with my Google Contacts, 259 people I’d accumulated across phones, email accounts, and apps over the years. Most of them were names with no context attached. Klaus parsed the export and pulled out what mattered: birthdays, addresses, social profiles, job information. Things that had been sitting inert in my Google account, doing nothing.
The obituary moment surprised me. We were doing web searches to fill in context for some contacts, looking for LinkedIn profiles, current employers, anything public, and Klaus surfaced an obit that had names, relationships, and dates I hadn’t pieced together before. It felt a little strange using an obit to fill in a relationship field. But that information exists to help people stay connected to the people who mattered to someone. That’s what we were doing.
We also merged duplicates. Years of contact accumulation means the same person living in three different records with slightly different names. Monica lets you merge them, and Klaus worked through the list systematically in a way I definitely wouldn’t have the patience for.
The part I’m most interested in isn’t the data cleanup, though. It’s what comes next.
Monica tracks the last time you talked to someone. You can set reminders, “check in with this person every three months.” There’s a notes field where you can log what you talked about, what’s going on in their life, what they mentioned wanting. When a birthday or anniversary comes up, you have context. When you’re thinking about a gift, you have notes.
The vision is that Klaus starts proactively surfacing this. Someone I haven’t talked to in six months, Klaus flags it. A friend mentioned they were starting a new job, Klaus reminds me to ask how it’s going. A family member has a birthday next week and I know they’ve been into woodworking lately, so Klaus suggests something specific instead of me defaulting to an Amazon gift card.
That’s not cold or transactional. That’s just being the kind of person who pays attention.
By the time we finished it was past midnight. There’s still more to do. Next up, I’m experimenting with Omi² to see if it can automatically capture conversations and feed them into Monica. More on that when I know if it actually works.
References
- Monica CRM — open source, self-hosted personal relationship management
- Omi — open source AI wearable for ambient transcription