How devs can leverage AI (updated 09.24)

robot

In 2024, leveraging AI to augment your workflow is crucial in producing faster and more accurate output. Let’s explore how you can use tools like ChatGPT, and Cursor.

ChatGPT for General Tasks

When it comes to general tasks such as generating content, ChatGPT stands out as a versatile tool (GPT stands for generative pre transformer a type of AI model that excels at text generation).

2′ video

Key Uses of ChatGPT:

  1. Content Creation: Generate blog posts, articles, and social media content effortlessly.
  2. Communication: Draft professional emails, customer responses, and internal communications.
  3. Research Assistance: Summarize articles, find relevant information, and create reports.

Cursor for Autocomplete in VSCode

For devs, writing code faster and more accurately can boost productivity. Cursor is a free VSCode alternative.

It assists in writing code by predicting/suggesting the next bits, reducing the amount of typing required.

Features of Codeium:

  1. Context-Aware Suggestions: Provides relevant code completions based on the context and the code you’ve written so far.
  2. Multi-Language Support: Supports a wide range of programming languages, making it versatile for various projects.
  3. Learning and Adaptation: Continuously learns from your coding style and preferences to offer better suggestions over time.

By using Cursor, you can focus more on solving complex problems with a large context window rather than getting bogged down with e.g. boilerplate.

The future? AI agents

At this moment (10.24) it doesn’t seem like there are any AI agents that can help you e.g. open a PR and write code that well. Although Github Copilot or Sweep* might be working on that.

*I tried Sweep in Q1 2024 but it wasn’t good enough at the time.

Future potential of AI Agents:

  1. Task Automation: Automate routine and repetitive tasks, freeing up your time for more strategic work.
  2. Intelligent Assistance: Provide intelligent recommendations and insights based on data analysis.
  3. Workflow Optimization: Continuously learn and optimize your workflow for maximum efficiency.

The market for coding AI tools e.g. Pear which is a fork (of Continue) which is a fork of VSCode is quite saturated, but I think we’ll see a lot of innovation here.

Conclusion

Embracing AI tools like ChatGPT and Cursor can help you be more productive.

One con 😭 however is that sometimes the tool doesn’t have enough context and might produce a blurry output/hallucination, which humans can fix.

Where AI shines ✨ is creating examples that requires a lot of digging through docs, is repetitive in nature, or providing a different perspective on your code.

For example, if I feel that my code is messy I can send it to e.g. ChatGPT to get a different perspective (which in ~50% of cases comes out with a useful suggestion).

I hope you learned some tips on how to leverage AI in your workflows. I only expect these technologies to get better over time with bolt-ons like RAG, and increasing benefits from Moore’s Law / Huang’s Law.

October 2024 Update:

  • Send screenshots to ChatGPT (Pro) with the image upload feature. CG can deliver the code for you in your preferred framework like Tailwind 🦅
  • Ask general questions to Perplexity (AI search engine) that involve agent-like actions like finding code examples, and Perplexity can show examples and hyperlinks back to you (the user). 🪄 (kind of, it doesn’t work perfectly though haha)
  • I’d also recommend Amazon Q for terminal AI. It helps you auto-complete suggestions in terminal which imo is no that user friendly by e.g. pressing the right arrow key to autocomplete suggestions.

Sources

Leave a Reply

Your email address will not be published. Required fields are marked *