Sat (Chennapragada Sathvik)

Tech, curious, and me
Building continuously.

[Sat (Chennapragada Sathvik) @SaSatChe]

Most supply chain dashboards are fundamentally broken. They tell you WHAT happened, stockouts, delays, margin drops, but they completely fail to tell you WHY.

​Enter Trita by Olynk AI: the first causal intelligence platform designed specifically for commerce operations.

[Sat (Chennapragada Sathvik) @SaSatChe]

Broke the application again. I asked my agent to add too many features at once. Damn it. Onboarding is a big deal, the first step and it's not working!

[Sat (Chennapragada Sathvik) @SaSatChe]

@12lijonathan @therealquippy @ycombinator Alright! Now that's exciting!

[Sat (Chennapragada Sathvik) @SaSatChe]

I asked someone for a favour, and now I feel like I am in debt to them. Why is that?

Did I lose my upper hand?

I am putting all my chips on the table for this. If it backfires, might as well. No regrets.

[Sat (Chennapragada Sathvik) @SaSatChe]

@gpusteve @MarathonMP @chemistry ahh comeon, now you're just trolling. This is great!

Wafers' no longer thin, huh?

Bad joke? Yeah, anyway, congrats, man!

[Sat (Chennapragada Sathvik) @SaSatChe]

Modi ji's got Aura 🔥

[MyGovIndia @mygovindia]

The Growth Story Just Hit 7.8%! 📈

7.8% GDP growth in Q1 FY 2026-27 marks a strong start for India and another boost to its growth journey.

PM @narendramodi captured the moment perfectly: “Doomsayers were doomed, and India bloomed... yet again!”

#GDPGrowth
#EconomicGrowth

[Sat (Chennapragada Sathvik) @SaSatChe]

@harrychow_ Okay, I checked about 10 hrs ago and this repo had 30 forks and around 15 stars. Now there are 70 stars and 178 forks!

Gotta give it to you, Absolutely insane strategy.

[Sat (Chennapragada Sathvik) @SaSatChe]

And of to sleep. Had a long day, very unproductive. That damn cold calls changed the course of today.

I am becoming too desperate for customers and maybe that's why I am not getting any?

Nahh, I have to analyse what I'm doing again. I need to get a building partner asap.

[Sat (Chennapragada Sathvik) @SaSatChe]

It's frustrating when no one lifts your call.

Ofcourse it's a cold call, ofcourse I'm a stranger and of course I'm selling my product.

I mean I still understand, I don't lift random calls but I expect the world to.

So LIFT MY CALLS.

[Sat (Chennapragada Sathvik) @SaSatChe]

@Atharv_Golait Yo, let's chat!

[Sat (Chennapragada Sathvik) @SaSatChe]

Tell me what's wrong.

[Atharv Golait @Atharv_Golait]

A 3rd-year college guy came up to me and said he’s a vibe coder.

“I know vibe coding very well.”

I asked how he builds a product.

“I copy-paste a Claude prompt into Claude Code.”

That’s the problem.

Building a SaaS is not “open Claude Code / Codex → drop a prompt → ship.”

I’ve been coding and shipping for 2 years. These tools make the process faster. They do not replace understanding.

Vibe coding is not prompt engineering.

It’s:
• security
• PRD
• markdown specs that actually constrain the system
• domain model + state machines
• database schema
• API spec
• how architecture maps to real functionality

If you don’t understand the code and why it exists, you don’t have a product. You have a demo that will break the first time a real user, edge case, or attacker shows up.

Understand first. Then build.

Tools speed up people who already know what they’re building. They don’t invent that knowledge for you.

[Sat (Chennapragada Sathvik) @SaSatChe]

Cold called a founder today as usual, for heavens the rudest guy I ever met.

Asked me if I didn't know it was festival today. Seriously?

Honestly I had the best calls today, people are happy and celebrating, until I called this guy.

Spoiled my mood for sure. Such an ass.

[Sat (Chennapragada Sathvik) @SaSatChe]

I don't think we should go further building deterministic rules for our predictions and numbers. It's not flexible enough to handle enterprise data.

Pivoting into building an agentic harness now, maybe we should focus on governance while using frontier LLMs rather than building the statistical rules and formulas.

[Sat (Chennapragada Sathvik) @SaSatChe]

@albinder @daily_objects Grofers doing B2B replenishment for DailyObjects vs a much larger range sitting on Blinkit now is a clean then-and-now. Listing is the easy screenshot. Who owns stock accuracy when the same bag is in a dark store and on the brand site?

[Sat (Chennapragada Sathvik) @SaSatChe]

Karnataka went after Amazon, BigBasket and Instamart on the licence, not just the seller, after Datura showed up listed as food. If the FSSAI number still renders on an unsafe SKU, the listing layer is the control. For a food brand on those shelves, who can actually pull it in the first hour?

[MediaNama @medianama]

Karnataka suspended the food licences of Amazon, BigBasket and Swiggy Instamart after finding poisonous Datura listed for sale as food, holding the platforms liable, not just the sellers.

When the signals meant to guarantee a purchase, ratings, badges, an FSSAI licence number, keep appearing on unsafe goods, can consumers still trust what platforms sell?

Join MediaNama's briefing call on Trust in Ecommerce.
📅 Sep 2, 2026
🕐 3:30–5:00 PM
📍 Virtual (Zoom)

[Sat (Chennapragada Sathvik) @SaSatChe]

@Atharv_Golait X is much better fs!

[Sat (Chennapragada Sathvik) @SaSatChe]

Does it really matter where you start? When the end obviously changes.

[Sat (Chennapragada Sathvik) @SaSatChe]

Got my first paper published in @IOPPublishing's Engineering Research Express: GAIOL, a framework for orchestrating multiple LLMs together.

Core idea: instead of fixed weights deciding which model to trust, we built an algorithm (ABTC) that learns per-task-domain reliability live, via Bayesian updates.

0.83 quality score, 24% above single-model baselines on a 500-query benchmark.

https://t.co/HfuO1PKyQY