Instead of measuring your team's obsession with customers (an input metric), focus on the output: building a product so compelling that customers become obsessed with you. As a coach tells players to score points, not just to sweat, the goal is the outcome, not the effort.
Factory refunded nearly $2M in revenue because their product wasn't creating "obsessed customers," even though they were skilled at sales. This painful, counterintuitive decision preserved trust, allowing them to re-engage those same customers later with a superior product.
Counterintuitively, a harness built to support multiple AI models is superior to one co-designed with a specific model. A multi-model approach prevents overfitting to one model's quirks, making the system more robust and higher-performing, analogous to how a model trained on the internet beats one trained on personal data.
To combat engineer skepticism, companies incentivized AI usage to the point of wastefulness ("token maxing"). This overcorrection is a faster way to achieve broad adoption than starting with strict controls. You can enforce responsible usage and cost efficiency once the value is proven and ingrained.
Open-weight models are often a generation behind the frontier. The correct comparison isn't against the latest proprietary model but its predecessor. On that basis, open models are now achieving parity, making them highly effective and cost-efficient for a majority of tasks.
The majority of AI-driven software development will shift from human-prompted (synchronous) tasks to autonomous agents working 24/7 (asynchronous). This "dark factory" concept means agents will identify, scope, and solve problems without direct, real-time human command, fundamentally changing the development lifecycle.
The core resource allocation question will evolve from budgeting for AI tools to choosing between hiring humans and buying compute tokens. Answering this requires a "software factory" with quantitative feedback loops to determine where each incremental dollar adds the most business value.
Factory's early DNA was forged by navigating a $5M valuation and giving back all its revenue. This "rock bottom" experience built deep bonds and resilience that companies anointed as unicorns from day one may lack, leaving them vulnerable during inevitable tough times.
Factory's vision for autonomous agents was correct, but the market wasn't ready for two years. This "time in the desert" highlights that market timing is as crucial as the idea itself. There are no consolation prizes for being early; you either succeed or you don't.