It was a surprise how huge the difference is when I switched from GitHub Copilot to Codeium. In practice it was a switch from using AI for repetitive tasks to using it for deep research during planning, better code understanding during bugfixing and real assistance in developing new features.
Real world use cases
This article focuses on some real world examples which I would like to share with you as creating one more list of Codeium’s key functionalities or one more list of its advantages over GitHub Copilot wouldn’t make a lot of sense.
Researching
We all can do that by ourselves. It’s not difficult, but just time consuming. Let’s try another way - ask a question and get the response based on analysis of multiple websites - an extract, no off-topic.
[P] Prompt: How much time does it take to read a cookie in Next.js?
[R] Response: Reading cookies in Next.js is generally very fast,
taking only microseconds to milliseconds of processing time.
The exact timing depends on a few factors:
1. Whether you're reading cookies on the client-side or server-side
2. The size and number of cookies being read
3. The specific method you're using to read cookies
In Next.js, you have several ways to read cookies:
Server-side:
(... skipped content ...)
Client-side:
(... skipped content ...)
The server-side methods are typically very efficient
since they're just accessing data that's already been parsed from the HTTP headers.
Would you like me to show you some examples of how to read cookies in Next.js,
particularly in the context of your server component?
You can search the web within the chat and continue the talk in context of the current codebase.
Refactoring
This is usually a repetitive task and it’s easy to get bored. Have you read how GitHub Copilot came with help for repetitive tasks? ( 🔗 GitHub Copilot for repetitive work ) Codeium does it even better, because it does not need exhaustive prompts, just a simple one is enough.
[P]: Apply the same changes to src/components/simple-product/form.tsx
as we did for src/components/configurable-product/form.tsx
[R]: I’ll help you apply similar changes to the simple product form.
First, let me examine both files to understand the changes we need to make.
Then it will present you a plan of all actions it will take if you approve.
Explaining
We often work with someone else’s code. Sometimes we can’t even ask the author when we have doubts about how some functionality works. Instead of slowly moving forward or getting stuck, we may ask a few questions to get a better understanding of the logic which stands for the functionality.
[P]: Describe validation rules used in shipping address
[R]: Validation Rules in ShippingAddress.tsx
Based on the code in the ShippingAddress component,
here are all the validation rules implemented using React Hook Form:
1. Company Name (shippingAddress.company)
- Required field with error message from translation key validation.requiredCompany
(... skipped content ...)
Estimating
It’s relatively easy to estimate a task when we already did something similar in the past although it may turn out difficult regarding new and complex functionalities. Splitting them up into chunks and asking your AI assistant how he would estimate some of these chunks may be a good starting point.
[P]: How much time would it take to set up Opensearch 2
in kubernetes and configure Magento for it?
[R]: Setting up OpenSearch 2 in Kubernetes and configuring it for Magento 2.4.6
involves several steps and considerations. Here's an estimate of the time required:
Estimated Timeline
Total time estimate: 8-16 hours
(depending on your Kubernetes expertise and whether you encounter any issues)
Breakdown:
1. Kubernetes OpenSearch Setup: 4-8 hours
(... skipped content ...)
2. Magento Configuration: 4-8 hours
(... skipped content ...)
Key Considerations
(... skipped content ...)
Final thoughts
Codeium is going to stay as it's useful in my everyday tasks. I found it helpful in areas I'm already familiar with, but be warned - as any other AI, it makes mistakes and it shouldn't replace strong knowledge gathering, problem analysis and planning. Let AI be your assistant, but still develop your skills.
đź’ˇ Prompt responsibly. An assistant can help with improving the quality of a product, but asking it to do your whole job for you is a bad idea.
