If you're deploying agents into a marketplace, a collaboration tool, or any product where value compounds with participation, you have a cold start problem — and wishing for traction is not a strategy.
Andrew Chen's The Cold Start Problem dismantles the mythology that network effects just happen. The real work is identifying the atomic network — the smallest cluster of people dense enough to generate genuine value for one another — and then solving for the hard side first: the suppliers, the creators, the skilled contributors whose presence makes the network worth joining at all. Growth is a sequence of deliberate engineering decisions, not a viral accident.
When agents handle execution, the cold start dynamic shifts but does not disappear. Your hard side is now whoever produces the signal your agents learn from or act on — early data contributors, power users whose behavior calibrates the system, domain experts who validate outputs. Seed that layer wrong and your agents optimize for the wrong thing from day one. The atomic network question becomes: what is the smallest human-plus-agent cluster that can demonstrate real value without outside dependencies? Identify that unit before you scale anything.
- Find the hard side of your agent network and recruit it deliberately before launch
- define your atomic network in human-plus-agent terms, not headcount alone
- escape velocity is a threshold you engineer toward, so instrument the specific moment your network becomes self-sustaining.
