top of page

RAG as a Service : The Developer’s New Best Friend

11/6/24

Source:

Alden Do Rosario for Artificial Intelligence in Plain English on Medium

Approaches

Why you don't want to build a RAG system.

Here’s the deal:


You would never ever in a million years build your own mail server — or CRM system — or in most cases, your own LLM.


Would you?


And yet, I see so many companies trying to build their own RAG pipelines.


Developers showing off their Langchain-powered RAG pipelines — without knowing the thousands of hours of pain that they are about to get into when they go to production.


No kidding — I just ran into a startup at a conference in Boston that literally launched their MVP using a Langchain powered RAG pipeline — and then days after — were sitting around yanking out Langchain from the tech stack.


This is terrible — instead of focusing their energy on product-market fit or other factors that would drive their business, they are now re-doing their tech stack — what a waste of developer energy — it’s almost like they are building their own LLM, rather than using the OpenAI API.


Here is a tip: RAG as a Service.

Latest News

12/3/24

Why Vertical AI Agents Could Be 10X Bigger Than SaaS: Insights from Y Combinator

Solution Trends

12/1/24

LLMOps Database

Case Studies

11/29/24

Enterprise AI Spending: What the Numbers Tell Us About Implementation

Analyst Research

bottom of page