An AI-powered read-it-later service that doesn't just store what you save — it surfaces what actually matters when you need it.
It is an AI-powered read-it-later service built end-to-end with Next.js 16 App Router, Prisma ORM, and Auth.js. Save articles, highlights, and links from anywhere — then let an AI layer surface what actually matters across your reading history.
No more forgotten bookmarks. No more overflowing reading lists with zero recall. It connects the dots across everything you've saved and makes your content library useful again.
Most knowledge workers have the same problem: they save articles, threads, and links constantly — but almost never go back to them. The content sits in a Pocket or Instapaper graveyard, never to be seen again.
Existing read-it-later tools treat saving and reading as two separate, disconnected acts. There's no layer that connects what you've saved with what you're currently working on or thinking about. The library grows, but it never gets smarter.
The insight: saving is only useful if you can reliably rediscover content at the moment it becomes relevant. That requires an AI layer — not just search.
The app adds a RAG (Retrieval-Augmented Generation) layer on top of a standard content library. When you save an article, it's parsed, chunked, and embedded into a vector store (Pinecone). When you ask a question or describe what you're working on, the AI retrieves semantically relevant content from your personal library and surfaces it.
The result: your saved content stops being an archive and starts being an active knowledge base that you can query in plain language.
The main interface lets you save, browse, and query your personal content library using natural language — powered by RAG over your own saved articles.
| Technology | Role & rationale |
|---|---|
| Next.js 16 | Full-stack framework using the App Router. Server Components keep the UI fast and the data-fetching simple; Route Handlers power the API layer without a separate backend. |
| TypeScript | End-to-end type safety across the data model, API routes, and UI components — critical when building on top of a relational schema with Prisma. |
| Prisma + PostgreSQL | Prisma ORM provides a type-safe database client generated from the schema. PostgreSQL stores users, articles, tags, and highlights with full relational integrity. |
| Auth.js | Authentication with email/password, Google OAuth, and GitHub OAuth — all with a single configuration. Session management and CSRF protection included out of the box. |
| Pinecone | Vector database for storing article embeddings. Chosen for its managed infrastructure — no self-hosting required, and it scales to millions of vectors without configuration overhead. |
| RAG pipeline | Articles are chunked and embedded on save. At query time, the top-k relevant chunks are retrieved and passed to an LLM with the user's question to generate a grounded, sourced answer. |
The core save-and-retrieve loop is functional. Auth, article storage, and the reader view are working. The RAG layer is integrated and returning relevant results from personal libraries.
I'm looking for Senior PM and Head of Product roles where product craft, technical depth, and user focus matter. Open to a conversation.