Raghavendra Prabhu (@omrvp) 's Twitter Profile
Raghavendra Prabhu

@omrvp

Cofounder & CEO, @TeziHQ

ID: 68922990

linkhttp://www.linkedin.com/in/rvprabhu/ calendar_today26-08-2009 07:22:38

1,1K Tweet

1,1K Takipçi

195 Takip Edilen

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

Me typing a response to a coworker question in Slack after spending hours tuning LLM prompts: "Think through it step by step as follows: (1)..."

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

A lot of the talk around improvement in language models is around capabilities e.g. 'now LLMs can do advanced math!'. I'd argue a more important vector to focus on is reliability i.e. what % of the time can the LLM accurately do what it says it does.

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

Many prospective founders I know try to get there rationally. This won't happen, probability weighted outcome is not favorable to the founding path. Rather, you have to feel an inner compulsion. A need to solve a particular problem.

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

There is a recent meme on AI models hitting a scaling limit. If we freeze LLM capabilities where they are now but got 10x/100x better in reliability and consistency, that would be more than sufficient to unlock incredible agentic applications.

Daniel (@growing_daniel) 's Twitter Profile Photo

Starting a company is awesome if you want to feel like a piece of shit one day and king of the universe next day and just keep alternating back and forth forever and ever

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

This post is the best summary I've seen of the key insights behind DeepSeek's success (I can say this having spent hours reading the relevant papers first). The main takeaway for the AI community should be the benefits of open research. All of this builds on top of other

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

"You can go off the rails completely when you are hyping yourself with the supply-side, versus really understanding how to translate that into real value to customers." Exactly. It takes a lot of skill and effort to take AI and deploy it towards a meaningfully differentiated

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

Conventional wisdom often defies common sense. I've wondered why that is, many times convention develops simply because one person did it that way at some point and everyone else followed suit since. Not always well thought out.

Raghavendra Prabhu (@omrvp) 's Twitter Profile Photo

Hiring a high quality team requires time commitment, but make sure you spend that wisely. With Tezi's AI agent Max, you can offload all the tedious tasks and focus on what matters most: talking to candidates to sell your vision and evaluate mutual fit. tezi.ai