Skip to content
AI-Native Portfolio

Building an AI-Native Portfolio with S3 Vectors

·1 min read
aiawss3-vectorsnext.jsrag

Why AI-Native?

Most developer portfolios are static sites. They work, but they don't leverage the content they contain. What if your portfolio could actually discuss your projects with visitors?

That's the idea behind this site. Every blog post and project page is embedded into Amazon S3 Vectors, enabling semantic search and RAG-powered chat across all content.

The Stack

The architecture is straightforward:

  • Next.js 15 with App Router for static generation + API routes
  • Amazon S3 Vectors for vector storage (up to 2B vectors per index)
  • Amazon Bedrock with Cohere for embeddings and Claude for chat
  • Vercel AI SDK for streaming chat UI

How Semantic Search Works

When you press Cmd+K, your query gets embedded via Bedrock Cohere, then S3 Vectors finds the most similar content chunks. No keyword matching — pure semantic understanding.

What I Learned

Building this taught me that AI features don't have to be complex or expensive. S3 Vectors costs 90% less than traditional vector databases, and Bedrock's pay-per-use model keeps costs under $20/month for a personal site.

The key insight: AI should enhance discovery, not replace browsing. The chat and search are there for visitors who know what they want. The traditional navigation is there for everyone else.

Related Content