Every few months, another property listing platform launches in Nairobi. AI has made it cheaper than ever. Yet if you ask people how they actually find homes, the answer is still WhatsApp groups, TikTok, and forwarded numbers.
We keep treating house hunting as a distribution problem. It hasn't been one for a long time. Listings are everywhere. The real friction is trust. Is the photo recent? Is the price accurate? Does the agent actually have access? Is the viewing fee standard or a scam? Is "Kileleshwa" actually past Ngong Road?
Most platforms try to remove the informal broker. That's a design error. In Nairobi, agents aren't a bug. They're the market interface. They hold the keys, know the landlords, move faster than any UI, and run on networks no centralized database can replicate. The problem isn't that they exist. The problem is there's no reliable way to verify them before you hand over cash.
So the question shifts. Not "how do we collect more listings?" but "how do we make the information people already see trustworthy?"
Building the interface is trivial. Search, contact, review. A weekend of coding. The hard part is the mechanic that decides whether the system survives contact with reality. An open review platform collapses immediately. Brokers flood it with fake praise, sabotage competitors, and drown out signal. You haven't built trust. You've built a louder noise machine.
The constraint has to be unforgiving: you can only leave a review if you initiated contact through the system. The platform logs the interaction. Forty-eight hours later, it prompts for feedback. Only that number can review, once. A verified badge attaches to the trail. This kills self-reviews, makes fake reviews expensive, and quietly turns the platform into the source of trust, not just a display layer.
The trade-off is volume. Fewer reviews. That's intentional. Ten verified interactions outweigh a thousand anonymous opinions. Signal over scale.
What this reveals isn't specific to rentals. It's a pattern across informal African markets. People aren't waiting for new behavior. They're waiting for better systems around the behavior that already exists. Whoever designs the right constraint in one market can port the playbook to others. The payment rails and language change. The trust mechanic doesn't.
House hunting doesn't need another listing platform. It needs a verification layer for the agent-led market that already runs the city. The technology is decided in minutes. The constraint is what decides whether it works.
I'm writing this down because the interesting part of building here isn't what ships. It's what you deliberately don't allow.