One thing I've realized recently is that serving frontier AI models is incredibly expensive. After experimenting with Ollama locally, I gained a new appreciation for the amount of compute cloud providers are handling behind the scenes.
While AI is becoming more affordable for developers, I'm taking the opportunity to explore what's possible with today's models.
My current side project is Factory—a headless AI orchestrator that coordinates multiple agents to build mini web apps from generated specifications.
Still very early, but it's exciting to see it starting to come together.

Top comments (0)