Effective working with documentation

0. Intro

SPIKE vs POC

Spikes and Proof of Concepts (PoCs) are both Agile techniques used to reduce uncertainty, but they differ in purpose: a Spike is a time-boxed investigation to gain knowledge or estimate effort, whereas a PoC is a, often throwaway, implementation designed to prove a specific technical concept or feature viability.

1. Task

User wants to buy a vintage turntable. Because everybody says he'll get basically better quality for less money than a modern one worth buying.
 

After checking the specialized forum he realized that it's not so easy, because they all have their issues and caveats, so he unsure anymore which one to buy.

The user has a personal AWS account, and he interested in saving costs, but also wants to be able to do everything in the cloud, without any local scripts or other overhead.

So what he needs is to build a simple RAG which will allow him to just send name of the specific model (for example: Dual 721), to the chat and get an answer if this turntable is considered good, how much issues this particular model has, any caveats, etc.

So, our requirements and constraints so far:

  1. Simple RAG
  2. Should be cost-efficient
  3. Forum to parse: https://www.winyl.net
  4. Everything should live in AWS Cloud

 

We need to estimate our SPIKE in days.

In other words - how much time we need to check related documentation*, and consider approach/approaches.

 

*Like what AWS services we need to use, how to ensure it's cost-efficient, and so on.

Let's play some poker!

2. Chatting (with AI)

Things can be build even a way more faster with coding tools ...

But I believe this approach can help in the next cases:

  1. You are not sure how exactly build what you need to build.
  2. When you need to build something with deeper understanding of all the components and why this is working this way.
  3. You want to document every step to share in with the others.

Disclaimer

3. Wrap up

We will continue by trying to actually build the app accroding to the guide which AI created for us. There could be also some eventual issues in this guide obviously, which we will try to mitigate.

On the next session:

Thank you!

Effective working with documentation

By Dzmitry Tsebruk

Effective working with documentation

  • 145