A colleague of mine left recently. She was one of the strongest connectors I’ve experienced in my career. The person who remembered what was discussed three meetings or five years ago. Who noticed when someone went quiet. Who made sure the right people talked to each other before things went sideways. She didn’t have a fancy title for any of this. It was just part of how she worked.
And now she’s gone. And you can start feeling it.
Not immediately. Not in a dramatic way. More like a slow creak in the structure. Things that used to just happen now don’t. Knowledge from years ago, not available at the ease of quick chat.
In woodworking, there’s a simple truth about glue: the strength of a joint depends on the surface area. More contact, stronger bond. Reduce the surface, and eventually the joint fails. Not with a bang, just with a quiet crack when you put weight on it.
I keep thinking about this in the context of AI.
We’re in a phase where companies are excited about AI taking over tasks. Summarizing meetings. Writing status updates. Distributing information. Answering questions that used to require a human to dig through documents and tribal knowledge. And honestly, AI is getting better at this.
But here’s what worries me. A lot of those tasks? They were the reason certain people talked to each other. They were the glue surfaces. The status update wasn’t just a status update. It was a moment of connection. The person who compiled it didn’t just copy-paste bullet points, they interpreted, they filtered, they added context that only someone embedded in the team would know. The act of gathering information was itself an act of relationship building. Now there are rumors that status reports are written by AI, sent to someone else where it gets summarized by AI and then an answer written by AI.
When we automate those tasks, we don’t just remove work. We remove contact surfaces. We shrink the area where human bonds form. And I’m not sure we understand what that means yet.
Can AI replace the mechanical part of this connective work? Absolutely. It can summarize, it can route information, it can even remind you that someone hasn’t spoken up in a while (probably). But can it read the room? Can it sense that two people are avoiding each other? Can it notice that a team is slowly losing trust, not because of a specific incident, but because of a thousand tiny missed moments of connection?
I don’t think so. Not today, and I’m skeptical about tomorrow.
And this is where it gets dangerous. Because the people who did this work, the human glue, they were often invisible. Their contribution didn’t show up in dashboards or sprint metrics. It showed up in the absence of problems. In things that just worked. In teams that somehow stayed aligned even when the roadmap changed every other week.
Here’s the cruel irony. Glue people spend a significant chunk of their time connecting others, bridging gaps, making sure the bigger picture stays coherent. That means their individual output, the stuff that shows up in a spreadsheet, is often lower. They’re not shipping fewer features because they’re lazy. They’re shipping fewer features because they’re busy making sure the whole team ships the right ones. But when layoffs happen, especially the large-scale kind where decisions are made from lists and numbers rather than individual understanding, guess who looks expendable? The person with the lower output. The person whose work doesn’t fit neatly into a metric. You let go of the glue, and you don’t even realize what you’ve done until the joints start failing.
When those people leave, you notice. When their tasks get automated, you might not notice at all. You just wake up one day and the team feels different. Slower. More brittle. More transactional. Less connected. And nobody can point to the moment it changed.
I don’t have a clean answer here. I genuinely don’t know how this plays out. But I think we need to be honest about what we’re doing when we rush to automate everything that looks like “coordination overhead.” Some of that overhead was load-bearing. Some of those inefficiencies were actually the thing holding the structure together.
Before you optimize the glue away, make sure you understand what it was holding.