AI transformation · Marketplaces · Growth
What a hackathon crawler taught me about where AI belongs
At Neighbor, our office was open plan. I sat next to the marketing team and right in front of our CPO and engineering, with data and enablement just one row over. That seating chart ended up shaping a lot of how I think about AI today.
Because when you sit in front of product and engineering, you hear everything. A new model came out, somebody already tested it. A new tool launched, someone was arguing about it by lunch. And for a marketer, that's dangerous in the best way. You start thinking, wait, can we use this for my problem?
My problem was hosts. Neighbor is a storage marketplace, and as Head of Host Marketing and Growth, I needed to find people with space to rent out. And not just anyone, we were looking for potential hosts with very specific characteristics. Doing that search manually took hours and hours. Someone would go through listings and pages, check criteria one by one, copy information into a sheet. It was slow, it was boring, and it was full of errors, because people get tired doing the same thing for hours.
So during a hackathon, I teamed up with our engineering team and we built crawlers to do that search. They identified potential hosts based on the characteristics we cared about. Hours of manual searching just went away.
And I want to tell you what I learned from that, because it wasn't only "automation saves time."
The best AI projects start with a painful, specific problem. We had a task that took hours, that nobody enjoyed, and that was slowing us down. When the problem is specific, it's easy to know if you solved it. When it's vague, you end up with a cool demo nobody uses.
Marketers should be in the room when things get built. A lot of companies still treat AI as an engineering or IT project. But I knew exactly what a good host looked like, which signals mattered, which ones were noise. Engineering knew how to build it. Neither of us could have done it alone, at least not that well and not that fast. If your business people only meet the technology at the launch presentation, you're missing half the intelligence.
The machine took the search, and the people kept the relationship. This is the part that stayed with me the most. The crawler found the hosts. But convincing someone to rent out their garage to a stranger is a human conversation. It's about trust, about answering their worries, about making them feel safe. When the team stopped spending hours searching, they had more time for exactly that. So the same team got to spend a lot more time on the part that actually needed them.
That's the line I try to draw in every company I work with now. AI is great at finding, sorting, matching, summarizing, spotting patterns across huge amounts of information. People are needed for understanding, persuading, caring, deciding what matters. When you put the work on the right side of that line, both sides get better.
Small and shipped beats big and planned. A hackathon has a deadline of a day or two. That forces you to build the smallest thing that works. I love that constraint. It's exactly how I run AI work now with traditional companies, through Fitch & Co: pick one painful task, build something small with the people who do it, measure, then grow. Big transformation plans look great on paper, and small wins are what actually change how people feel about the technology.
And proximity matters. I don't think I would have thought about crawlers if I didn't sit in front of product and engineering every day. Ideas came from overhearing conversations, from asking "how hard would it be to..." over the desk. A lot of traditional companies have their tech people on another floor, in another building, sometimes in another company. If you want AI ideas from the business side, put the business close to the people who build. Even one day a week sitting together changes things.
Today when a leader asks me where AI belongs in their company, I think about that crawler. Look for the work that's searching, sorting and copying, the hours people would love to give away. Give that to the machine. Then look at what your people would do with those hours, and protect that part with everything you have.