AI product development involves navigating fundamental trade-offs between competing priorities. A framework borrowed from project management theory - the iron triangle - provides a structured way to analyze these tensions. The framework examines how scope, cost, and time interact at design-time, and how quality, cost, and latency interact at run-time.

At design-time, AI teams face constraints between feature scope, development costs, and time to market. The relationship follows a basic mathematical model where development cost equals scope divided by time, multiplied by a productivity factor. For instance, building a 300-story point AI product in 100 days with a productivity factor of 0.012 would require approximately $125,000 at $500 per story point. This model helps teams understand that stretching timelines or reducing scope can lower costs.

The run-time iron triangle focuses on customer-facing trade-offs. AI products must balance response quality, inference costs, and latency. Higher quality responses delivered with lower latency cost more to produce. A system achieving 90% accuracy with 0.5-second response times will have different cost implications than one with 70% accuracy and 2-second responses. Understanding these trade-offs helps product teams make informed decisions about system architecture and deployment strategies.