Focus on Loops Not Lanes

We're entering an era where products evolve from static artifacts to living organisms that continuously learn and improve through interaction. This shift fundamentally changes how we build and measure success in technology.

The traditional approach of shipping a product and making incremental improvements is giving way to products with high metabolic rates—systems that ingest data, digest rewards models, and continuously optimize for specific outcomes. As Asha explains, "All of a sudden these are these living organisms that just get better with the more interactions that happen... I think this is the new IP of every single company."

This evolution demands a different mindset from builders. Rather than focusing on functional lanes (PM, engineering, design), successful teams obsess over the entire feedback loop. The metabolism of a product team—how quickly it can ingest data, learn from it, and improve—becomes the competitive advantage. Teams that can rapidly iterate through this cycle gain velocity and throughput that traditional organizations can't match.

For leaders, this means organizing around loops rather than lanes. The traditional organizational chart with clear functional boundaries becomes less relevant as cross-functional capabilities become essential. Teams need to be structured for continuous feedback, with observability becoming central to the culture.

For individual contributors, this shift requires developing polymath capabilities. The most valuable builders will understand not just their discipline but the entire product creation cycle—from cost and performance considerations to rewards design and user experience. As Asha notes, "I think we're seeing this advent of the polymath... where full stack builders are having their renaissance."

The implications are profound for planning and strategy. Rather than rigid roadmaps, teams need to organize around "seasons"—periods defined by secular changes in the industry or customer needs. These seasons might last three months or a year, but they provide the shared context for quarterly OKRs and shorter-term goals, with slack deliberately built in for adaptation.

In this new world, the products that win aren't necessarily those with the most features, but those with the strongest learning loops—products that think, live, and learn.

The Rise of Post-Training

A critical shift is happening in AI development—the move from pre-training to post-training optimization. As models reach certain parameter thresholds (around 30 billion), it becomes more economical to fine-tune existing models than to train new ones from scratch.

This creates opportunities for companies to differentiate through how they adapt models to specific use cases rather than building foundation models. Teams should invest in reinforcement learning capabilities and the infrastructure to continuously improve models based on real-world usage and feedback.

The Agentic Future

As we approach a world where "the marginal cost of good output is approaching zero," we'll see exponential demand for productivity that can only be met through agents. This will transform organizational structures from traditional hierarchies to work-centered networks where tasks and throughput become more important than reporting relationships.

For leaders and ICs alike, this means developing skills to work effectively with and manage agents—setting clear objectives, reviewing outputs, and continuously improving agent performance through fine-tuning and observability.