jnaapti

Startup of @gauthampai | Building services and agents for learning, training and personal knowledge management

[jnaapti @jnaapti]

Planning a cohort starting mid-June. DM if interested.

Course:
Agentic Engineering: Rethinking Knowledge Work
From ad-hoc AI usage to structured agentic workflows

Course timing will be favorable for people between GMT to IST timezones.

[jnaapti @jnaapti]

15 years ago, today, I registered the jnaapti domains and created this account!

[jnaapti @jnaapti]

@SinghSaksham01 @gauthampait You got the wrong @gauthampai 😄

[jnaapti @jnaapti]

The next gen Virtual Coach😍

[jnaapti @jnaapti]

@shunnode Did you try Gemini in GitHub Copilot or elsewhere?

[jnaapti @jnaapti]

Has anyone noticed this behavior where if you continue a conversation for a very long time Codex starts getting lazy and when debugging issues says things like "the likely cause is ..." without investigating it.

[jnaapti @jnaapti]

@shunnode I have been active in my @gauthampai account for a few days now. 😄

Been busy hatching some product ideas and talking about it as I build it.

[jnaapti @jnaapti]

Built a beautiful Notes Calendar in 10 minutes with Grok Code Fast 1 (Preview).

#buildinpublic

[jnaapti @jnaapti]

Link to the original article: https://t.co/RdlSU3zrRY

[jnaapti @jnaapti]

The big issue I have with AI Coding agents is exactly this. The longer the code generated by AI, the more time it takes to fix the "last mile" issues.

And yeah, AI can game tests so that doesn't help.

[jnaapti @jnaapti]

Do you remember when you joined X? I do! #MyXAnniversary

[jnaapti @jnaapti]

Before the advent of LLMs, machine learning had already made significant progress. We had summarizers, classification/clustering algorithms, and other solutions that were about 80% effective. However, the lack of final-mile refinement often kept them from being production-ready. With LLMs now in the picture, many of these use cases are resurfacing.

However, instead of leveraging traditional ML approaches where they still work well, many have jumped straight to LLMs for everything. In reality, traditional ML solutions can serve as an effective first-level filter - delivering solid results while keeping time and costs under control. I've found this approach to be incredibly practical, yet seems overlooked. Why aren’t more people talking about this?

[jnaapti @jnaapti]

Will be off the blue tick soon. I will enable it if and when I feel like it. But verified or not, it's me!

[jnaapti @jnaapti]

#DeepThink (R1)

Note the time taken

[jnaapti @jnaapti]

Here is a "Deep Thinking" question:

There are 2 companies. One offers a 10% flat discount on my purchases. If I buy ₹100/- worth of items, they will charge me ₹90/-. Another offers a 15% discount but they give the discount as 15 points for every ₹100/- spent. I can then redeem these at the rate of ₹1 per point. However, when I redeem, they don't offer discount on the redeemed points. Which one offers a better discount over time?

Let me know how different AI models answer it.

[jnaapti @jnaapti]

#ChatGPT4o