Paths to outcome: A new model to architect platforms

Platforms are often described vaguely, and that vagueness is expensive. This session introduces the Paths to Outcome model: a framework for describing platform architecture with enough precision to make disagreements productive.
Kaspar defines the core unit of any platform, the path, as a repeatable route from intent to outcome, and walks through its components: intent, interface, capability, and output. He unpacks what makes a capability well-designed: a stable promise with hidden internals, decoupled from the tools that implement it and free to evolve underneath. He then shows how paths compose into platforms alongside value streams, with identity, policy, and state as the cross-cutting concerns that make the model hold in practice.
Finally, he applies the model to agents, showing why agents need paths too and why good platform design is what separates productive agentic loops from unconstrained hallucination at scale.