Last Wednesday night, I stared at Berlin’s event list and felt strangely empty

At 9 p.m., I was scanning next week’s meetups from a café in Kreuzberg. More than twenty events appeared at once: design, AI, startups, each with a polished description. I scrolled for ten minutes and still had no idea which one I should attend. That was when I realized that the more AI wrote, the more I needed someone—or something—to help me decide.

What I finally understood: the missing piece wasn’t events. It was the person filtering them

All About Berlin is a site that brings the city’s events together and uses AI to process large amounts of information. My colleague Lena used to find events through it, but she had a habit: before deciding whether to go, she’d ask herself, “Can I actually talk to people there?”

On a Thursday in April 2025, outside an AI meetup in Mitte, a stranger’s recommendation led Lena to the collaboration client she later worked with. She said the real time-saver wasn’t reading the full list. It was narrowing the list first.

Here’s what I do now: I ask AI to narrow the events down to three based on budget, distance, topic, and attendee type. Then I read the host’s description myself and choose one to attend. The machine moves the information; the human keeps the judgment, chance encounters, and relationships. I’m not letting AI take over socializing—I’m putting it where it actually belongs.

It costs little to copy, but don’t automate everything on day one

The setup costs roughly RMB 0–30 per month and can run on free web pages or an AI assistant I already use. Initial configuration takes about 20 minutes. Anyone comfortable copying and pasting information can do it.

The main technical challenge is avoiding vague filtering prompts. Instead of “find me events,” I use criteria like: “Under €20, a 20-minute walk from my office, no more than eight attendees, and open to on-site conversation.”

For the first step, I open the AI chat tool I already use, start a new conversation, paste in the event-page link and those criteria, and let it organize three candidates. I keep the manual review step for now; there’s no need to rush into connecting website data or building an automated system.

If I were starting from one of these three stages, this is how I’d begin

If I had just started out, I’d manually screen ten events, attend only one, and focus on meeting people before building any process.

If I already had one or two clients, I’d add their needs to my filtering criteria and keep only three events each time. Otherwise, I might waste hours on events that looked lively but offered no real business conversations.

If I was scaling up, I’d have the team record “event—number of people met—follow-up opportunities” in a consistent format, while keeping a personal reflection after every meetup.

This approach isn’t for everyone, and it’s useless for businesses with no connection to AI. There’s no pressure to try it now either—I can start by choosing the next event I genuinely want to attend.