Run Your Harper AI Agent on Google Cloud Vertex AI — 3 Files Changed

In Part 1 we built a conversational AI agent on Harper — semantic cache, vector memory, local embeddings, web search, chat UI — all in one process. It talked to Claude through Anthropic's direct API. That works great for solo developers and startups. But if your org runs on Google Cloud, you probably want Claude going through Vertex AI — same billing, same IAM, same audit logs as everything else in your GCP project. Good news: it took three file changes and zero rewrites to the agent logic....