Last week, @vercel.bsky.social announced their AI SDK 3.0 with support for custom LLM tools and per-tool rendering. It might not be obvious that "tools" make for great RAG apps - here's how:
dewykb.github.io/blog/rag-too...

Typescript is a perfect fit for your RAG app | Dewy Why TypeScript/JavaScript excels in RAG app development, offering rich prompt engineering, seamless API integration, and dynamic UIs for modern web apps.

What do you think - are there good reasons *not* to use Typescript/Javascript to build RAG applications these days?

dewykb.github.io/blog/typescr...

Some collected insights from the RAG apps I've helped put into production, and which have informed the design of Dewy: dewykb.github.io/blog/rag-for...

Building a question-answering CLI with LangChain.js and Dewy | Dewy This guide will walk you through building a question-answering CLI using LangChain.js for prompting, the OpenAI API for the language model, and Dewy as your knowledge base.

Yes, Dewy works with LangChain too (and makes it a lot easier to build Gen AI apps using your docs)

dewykb.github.io/blog/qa-cli-...

Building a RAG chatbot with NextJS, OpenAI & Dewy | Dewy This guide will walk you through building a RAG application using NextJS for the web framework, the OpenAI API for the language model, and Dewy as your knowledge base.

First post and it's shilling for an OSS project I'm working on. Let's get this party started.

dewykb.github.io/blog/rag-app...