What scaling teams get wrong about automation

The flashy-project trap

When Peter and his co-founders started Segment, they weren’t building an analytics company. They were building a classroom tool nobody wanted. What turned it around wasn’t a grand idea. It was a small, repeated request they nearly ignored.

“We kept hearing the same thing: ‘Can we just use your data pipeline?’ We almost waved it off, because it wasn’t the product. Turned out it was the only thing anyone actually wanted.”

He sees the same thing with AI now. Teams chase the impressive build and walk straight past the dull, repetitive problem in front of them. The one that would quietly hand everyone back an afternoon a week.

Why the boring wins get missed

“The valuable problem is usually unglamorous. Nobody gets promoted for automating the thing everyone hates doing. So it just sits there, costing you, year after year.”

His advice is refreshingly simple: watch for the work people quietly route around. The spreadsheet someone rebuilds every Monday. The approval that always takes three days. That’s usually where AI earns its keep first.

Catch the idea before it evaporates

“You’ll have a conversation, someone names the real problem, and two weeks later nobody can remember what it was. The insight just disappears.”

That’s the gap we try to close: turning a half-remembered “we should really fix that” into a project with a number attached and a date on the calendar.

The long game

“The best opportunities don’t announce themselves. They whisper. Your job is to build something: a habit, a tool, a team that catches the whisper before it fades.”

Thanks to Peter for the conversation. More Bronn Conversations soon.