How I use different LLMs for writing, coding and research (So you can improve your productivity and results)

Do you like like my art?

AI transformed my work life last year. I went from spending 8 hours writing a blog post to just 2 hours. Like most people, I ONLY used ChatGPT and thought it would do it all….

But here’s the thing – not all AI tools are created equal. I learned to use different LLMs for different tasks for optimal results.

I’ve spent countless hours testing ChatGPT, Claude, and Gemini. I’ve used them for everything from writing code to market research. Now I know exactly which tool to use when. And I’m going to show you how to do the same without spending 100+ hours.

Writing: Picking the Perfect AI Partner

Most LLMs write generic content you can tell that it’s A.I. Nobody likes reading that. After months of testing, here’s what I learned:

GPT-4 shines at creative writing and storytelling. It captures voice and style better than other AIs. I use it for:
• First drafts of blog posts
• Creative brainstorming
• Generating fresh angles on topics

Claude excels at structured content and research. Its outputs are more factual and organized. I rely on it for:
• Creating detailed outlines
• Summarizing complex topics
• Editing and fact-checking

Gemini works best for quick, informal writing. It’s great at:
• Social media posts
• Email responses
• Short-form content

My choice

I like Claude the most. I don’t have any formal study but I just personally think Claude gives the most human like writing style

Coding: Your AI Programming Buddy

I’m not an expert coder but I use A.I. to write alot of mini apps like a wordpress quiz that connects to a pocketbase database. I also copy and paste all the error messages with my self hosted apps and wordpress into A.I. to help me figure out how to get it running.

Cursor is my go-to for really complicated deep coding, not so much for a quick web app. It’s like having a senior developer watching over your shoulder. Benefits include:
• Apply changes to code instantly
• Code editor interface without copy and pasting
• A full blown app that has complexity

ChatGPT his great for quick prototypes because it has a preview window and you can see what the app looks like. It’s perfect for:

• Quick prototypes
• Simple web apps
• Debugging problems

Claude writes better documentation than any other AI. It explains code clearly and adds helpful examples. I use it to:
• Document functions
• Write API guides
• Create code comments

My choice

For my use case, I just chatGPT the most. I just love how easy it is to see a preview and export it. When I need more complicated stuff, I go to cursor.

Research: AI-Powered Discovery

Most LLMs don’t search the internet for current data. It stores all the data in the model so it gets outdated for current stuff. For tasks where I want to find current information such as what’s the current favorite candidate to win the US elections, I’ll need something that does search the internet.

For this one, there’s 1 choice: Perplexity.

Free LLMs

Most people use the free tier of all these LLMs and it’s fine. For power users like myself and especially when you use automations, the A.I. costs can really rack up fast. I paid $300 last month and it shocked me.

I use Meta’s LLAMA for alot of the grunt work stuff that’s short. Like summarizing things, finding keywords.

For image recognition, I use Google Gemini. I use this to identify home items by uploading images of them into my homebox.

AI isn’t one-size-fits-all – I learned this the hard way after wasting months using ChatGPT for everything. Each AI model has its sweet spot, and knowing which to use when has doubled my productivity.

I spend my days writing content, coding projects, and doing deep research. Through trial and error, I’ve found the perfect AI tool for each task. I’ll show you exactly how I use different AIs to get better results in less time.

By the end of this article, you’ll know which AI to use for writing, coding, and research tasks. Plus, you’ll get my tested prompts and workflows that took me months to perfect.

My Daily AI Workflow

Here’s my exact process:

1. Morning writing with GPT-4
2. Code assistance from GitHub Copilot
3. Research with Claude and Bard
4. Final editing with Claude

This combination saves me 4+ hours every day. The key is switching tools based on the task.

What’s Next?

Start with one AI model you’ll use daily. Track your results. Join communities where others share AI tips.

The tools keep improving. But the basic idea stays the same – pick the right AI for each task. Your productivity will thank you.

Want to learn more? Join my free AI productivity newsletter. I share new tips and workflows every week.