-
Hey there! Welcome to TFC! View fewer ads on the website just by signing up on TF Community.
How do you actually use AI tools in daily life and projects?
But you can tell the underlying model, no?
I've used pro of both and chatgpt feels better to me
better than chatgpt.But claude sonet 4.0 thinking model with cursor is the best so far imo
First of all, AI tools compilation thread link here.
I am not into coding, so I don't use AI tools for coding, but in my company I have heard developers using copilot, cursor, windsurf etc.
I use gemini, chatgpt, and perplexity.
Gemini - I got the 2TB AI student plan. helpful in composing emails in gmail. just write what i want to in the prompt that appear within gmail compose box and it will write mail. Then sentence restructure, correction etc. it also shows email summary etc.
chatgpt - is better, I upload pdfs and ask to analyze, summarize, write email based on some data from pdf, things like that. I don't have the paid version. just using the free version.
perplexity - I was using it for searching results from web and providing answers, but sometimes replies are not correct (like reward rate of credit cards and all are wrong, it takes data from some cardmaven etc, not fully reliable). Again, not paid, free version.
Apart from this, I haven't used many tools, still lagging a lot beind in AI adoption.
I am not into coding, so I don't use AI tools for coding, but in my company I have heard developers using copilot, cursor, windsurf etc.
I use gemini, chatgpt, and perplexity.
Gemini - I got the 2TB AI student plan. helpful in composing emails in gmail. just write what i want to in the prompt that appear within gmail compose box and it will write mail. Then sentence restructure, correction etc. it also shows email summary etc.
chatgpt - is better, I upload pdfs and ask to analyze, summarize, write email based on some data from pdf, things like that. I don't have the paid version. just using the free version.
perplexity - I was using it for searching results from web and providing answers, but sometimes replies are not correct (like reward rate of credit cards and all are wrong, it takes data from some cardmaven etc, not fully reliable). Again, not paid, free version.
Apart from this, I haven't used many tools, still lagging a lot beind in AI adoption.
Is the company promoting, or are the devs proactive
How did you manage to the 2TB AI student plan?
My Fixed tools are ChatGPT and Cursor.
I am dabbing with Claude (specifically for Claude Code to see how efficient it is and is it really better than just using Claude model on Cursor).
For company, I use Devin for work, sometimes Copilot. (Not so much).
Airtel gave Perplexity for free - So instead of basic Google search, I use that. Sometimes that and sometimes Google AI Mode.
I am dabbing with Claude (specifically for Claude Code to see how efficient it is and is it really better than just using Claude model on Cursor).
For company, I use Devin for work, sometimes Copilot. (Not so much).
Airtel gave Perplexity for free - So instead of basic Google search, I use that. Sometimes that and sometimes Google AI Mode.
Rare to find such companies that are bullish on AI!
In India, employment costs are cheaper. (My perspective is with respective to CS/IT related jobs). So, for businesses, they'd rather have a higher headcount than to pay for AI tools. (Plus major companies are Services sector where hourly billing is done per employee, so having AI tools kind of eats into their profits I believe).
yes but companies providing tools to use is a different ballgame. I like these companies; leverage them to upskill as well. win win
Mine did a huge poc
We got all the tools did a big comparison between all collated everyone's feedback and decided on which to keep
Next up is cli ai tools
Our entire leadership and career pathing, hiring
Everything has been redone to include ai and ai tools and innovations
We're constantly pushed to use ai for as many use cases as possible and it is now a personal/team and company metric
That x items are mandatory that they should have ai involvement etc
Each model and their purpose: Personally do not like Claude and Gemini, I'm not able to sync with them, so much gap between my requirement and their understanding.
docs.github.com
AI model comparison - GitHub Docs
Compare available AI models in Copilot Chat and choose the best model for your task.
docs.github.com
| Model | Task Area | Excels At (Primary Use Case) |
|---|---|---|
| GPT-4.1 | General-purpose coding & writing | Fast, accurate code completions and explanations |
| GPT-5 mini | Deep reasoning & debugging | Well-defined tasks and precise prompts |
| GPT-5 | Deep reasoning & debugging | Multi-step problem solving and architecture-level code analysis |
| o3 | Deep reasoning & debugging | Multi-step problem solving and architecture-level code analysis |
| o4-mini | Fast help with simple/repetitive tasks | Fast, reliable answers to lightweight coding questions |
| Claude Opus 4.1 | Deep reasoning & debugging | Complex problem-solving challenges, sophisticated reasoning |
| Claude Opus 4 | Deep reasoning & debugging | Complex problem-solving challenges, sophisticated reasoning |
| Claude Sonnet 3.5 | Fast help with simple/repetitive tasks | Quick responses for code, syntax, and documentation |
| Claude Sonnet 3.7 | Deep reasoning & debugging | Structured reasoning across large, complex codebases |
| Claude Sonnet 4 | Deep reasoning & debugging | Performance and practicality; perfectly balanced for coding flows |
| Gemini 2.5 Pro | Deep reasoning & debugging | Complex code generation, debugging, and research workflows |
| Gemini 2.0 Flash | Working with visuals (diagrams/screenshots) | Real-time visual reasoning for UI and diagram-based tasks |
| Grok Code Fast 1 | General-purpose coding & writing | Fast, accurate code completions and explanations |
We decided to keep
Chat gpt enterprise license with codex
Copilot
And cursor
And some open source agentic stuff
We're also considering claude code and other cli tools very soon
Claude code is good for PR Reviews directly. They have their github action (assuming you use GH) where you can just mention claude and ask to do review / changes / etc.
That's what a frontier model can do. But this thread is about how one is using in their day-day personal/professional life. i.e. ai agents , frameworks etc.
Kushagra, m4b gave good insights.
Currently many are struggling to adopt GenAI and ending up like IDFC CC. https://www.technofino.in/community/threads/idfc-cards-facelift.35851/post-888250
Codex also has pr reviews for GitHub
Same process
It's also proving out to be nice
Feedback so far is some people are rating it higher than claude
Yet to use codex. If you want you can build your own PR reviewer agent.
An example: https://docs.cline.bot/features/slash-commands/workflows
You can slowly structure it out; it will be efficient.
No such action required, you can add this prompt to a file and request AI to use it. Only calls gh cli for GitHub APIs. Anyway, better let the security team know about this.
