Rohit Mundada

Agentic engineering @moengage

[Rohit Mundada @_mrohit]

@championswimmer It's always the same smell test to identify AI generated posts: "That's not X, that's Y"

[Kartik Kannan @kartik_kannan]

Have never found Bellandur a great place to live. While I used to work in Jio (Bellandur Junction), I preferred staying in an area that was a proper place (J P Nagar) with access and proximity to places of culture, restaurants, theatre, parks etc. Bellandur and ORR is just a concretised place full of traffic jams on ordinary days and bad traffic jams when it rains.

[Rohit Mundada @_mrohit]

@sandygrains My wife has her own studio in Bangalore. She has been working extensively on similar projects in Bangalore and Mumbai.
You can have a look here: https://t.co/GEVv33tPHf

[Rohit Mundada @_mrohit]

@dharmeshba Agree. I would not make any big ticket and cognitive load purchases like Life and Health Insurance from a "UPI app".
But I might still quickly renew my bike/car/travel insurance from these apps just because they reduce the friction and the cognitive cost is cheaper.

[Rohit Mundada @_mrohit]

“Cheat on everything”

→ pivoted to enterprise SaaS faster than anyone expected 😭

PS: still one of the best AI marketing runs I’ve seen!

[Cluely @cluely]

hello world

[Rohit Mundada @_mrohit]

The funniest thing about AI startups:

They begin with
“this changes human civilization forever”
…and 18 months later they’re optimizing sales calls and meeting summaries for enterprise SaaS 😭

[Rohit Mundada @_mrohit]

The jump in speed for ChatGPT 5.5 is insane!!!

Blazingly fast to the point where it feels like it’s not even processing, yet the quality is still hitting harder than Claude or Gemini. Anyone else finding it hard to go back to "slow" models now?

What a time to be alive!

[Rohit Mundada @_mrohit]

@harshilmathur It feels like some things are not adding up. breakthroughs like this would have been published in some top conference, not a launch video.

[Rohit Mundada @_mrohit]

@system_monarch It also depends what "model" means here, most of the "model" interactions out there are already an abstracted layer above base LLMs.
This will involve system prompts, thinking tokens.
Tool calls will come into picture if the "model" has capability to invoke.

[Rohit Mundada @_mrohit]

@banglani I’m in, and also looking to solve the day to day problem of similar people about managing their money using AI tools

[Rohit Mundada @_mrohit]

@anmolm_ @Ajain112 @shobhitic That sums up why Ycom chose BLR

[Raphael Schaad @raphaelschaad]

I don't think we use AI for too much. I do think we overuse AI too deep in the stack.

The mistake is routing work through LLMs when deterministic compute would do it faster and more reliably.

Take email triage. One approach is to send every incoming email through a gigantic model, consult markdown skills, update memory, and let the model decide what to do every single time. A better approach is to use AI to study the inbox, detect patterns, and generate/manage rules. Then let deterministic filters do the repetitive work instantly. Archive it. Label it. Move it. No model call needed.

This is true for agents, and honestly for humans too: pilots and surgeons handle repetitive tasks best when there's a clear process.

The most interesting systems are not AI all the way down. They use AI where intelligence is needed, then hand off to deterministic compute where possible.

AI decides the rules. Deterministic systems run them.

[Teknium 🪽 @Teknium]

Want to do this in Hermes Agent? We've had support for this since early March, check out the docs on setting it up, with no 5 per day limits, on your own server, and with any model!

See the guide here:
https://t.co/NYWybGqnxz

[Rohit Mundada @_mrohit]

@rohitkamathh @waitin4agi_ Totally agree the fitness movement and those emotional moments are powerful and positive, but none of that really justifies the ₹9k price tag for most people, does it? What specifically makes the cost feel worth it beyond the hype and validation?

[Arnav Gupta @championswimmer]

No this is a "manager brain" problem (saying this, yet as a current holder of the same job title).

Manager brain: "if me who doesn't code hands on can make Claude do it 90% right in 5 min, then the IC should be able to do it 100% right in 5 min"

[Rohit Mundada @_mrohit]

@asxna Okay, am I overthinking this or is it really sarcasm?

[Rohit Mundada @_mrohit]

@xlampsx @bcherny @doodlestein I have tried resetting the session, restarting my Mac, and reinstalling Claude code. Nothing worked.

[Rohit Mundada @_mrohit]

Laughable at best: An AI company building the future of work, yet their own support agent is a "broken record" bot that can't solve a 4-day outage. If this is the "replacement" for knowledge work, we’re in trouble.

[Rohit Mundada @_mrohit]

@bcherny @doodlestein It’s not even a rate limit issue, I haven't had a single response from Claude Code in 4 days. For a paid product, the customer support is abysmal. All the AI agent does is spam standard RL documents. This isn't what I’m paying for.

[Rohit Mundada @_mrohit]

@bcherny @doodlestein It’s not even a rate limit issue, I haven't had a single response from Claude Code in 4 days. For a paid product, the customer support is abysmal. All the AI agent does is spam standard RL documents. This isn't what I’m paying for.

[Wise @trikcode]

the generation that refused to accept cookies.

is now giving AI access to their desktops, files, and bank accounts.