2025 Marketplace Wrap-up, Boatsetter/Getmyboat and Coursera/Udemy Mergers, Coinbase and Kalshi Partner, Shopify and Uber Integration, Grindr All in on AI, Whop Powers mirco1 Payments and Much More!
This read well as a map of coexistence rather than disruption, whereAI marketplaces scaling violently, legacy models like Craigslist still working, and network effects quietly doing what they’ve always done.
It made me think about how many of the breakout cases you highlighted weren’t “better marketplaces,” but marketplaces that unlocked new economic surfaces - evaluation labor, prediction, agent-led workflows, where price sensitivity and norms hadn’t yet hardened.
I’m curious whether you see a point where agents start flattening differentiation across marketplaces, pushing defensibility away from matching and toward who owns reputation, workflow, or capital rails over time?
The Mercor trajectory is wild but the underlyng thesis about AI evaluation labor makes total sense. Trained an NLP model last year and ended up using about 40 contractors to label edge cases, and the quality gap between domain experts and general labelers was huge. The 4658% growth isn't jsut about scaling a platform, it's about timing the exact moment when every lab is scrambling to teach models nuance in specialized domains.
Thanks for this!
Thanks for sharing this Colin!
This read well as a map of coexistence rather than disruption, whereAI marketplaces scaling violently, legacy models like Craigslist still working, and network effects quietly doing what they’ve always done.
It made me think about how many of the breakout cases you highlighted weren’t “better marketplaces,” but marketplaces that unlocked new economic surfaces - evaluation labor, prediction, agent-led workflows, where price sensitivity and norms hadn’t yet hardened.
I’m curious whether you see a point where agents start flattening differentiation across marketplaces, pushing defensibility away from matching and toward who owns reputation, workflow, or capital rails over time?
The Mercor trajectory is wild but the underlyng thesis about AI evaluation labor makes total sense. Trained an NLP model last year and ended up using about 40 contractors to label edge cases, and the quality gap between domain experts and general labelers was huge. The 4658% growth isn't jsut about scaling a platform, it's about timing the exact moment when every lab is scrambling to teach models nuance in specialized domains.