Back to Blog

Home / Blog / Why I Started Building Agentic AI Systems

Machine LearningCareerPublished

Why I Started Building Agentic AI Systems

November 15, 20254 min read

How a side project turned into a deeper interest in multi-agent architectures, and what surprised me along the way.

How It Started

A small side project to automate financial research turned into a much deeper interest in how multiple specialized agents can collaborate to solve a task none of them could handle alone.

What Surprised Me

  • Giving an agent web search access changed the quality of its reasoning, not just its access to facts
  • Coordinating multiple agents is closer to systems design than traditional ML work
  • The hardest part wasn't the AI — it was deciding when to trust the agent's output versus verifying it

Tags

Agentic AILLM