Jagannath Putrevu

@OpenAIDevs embeddings api fix when? Outage going on for a while now

I'm Boris and I created Claude Code. I wanted to quickly share a few tips for using Claude Code, sourced directly from the Claude Code team. The way the team uses Claude is different than how I use it. Remember: there is no one right way to use Claude Code -- everyones' setup is different. You should experiment to see what works for you!

Feels like a glimpse into the future of operating systems. All I can think of is how a phone that is not build around apps but built around actions could look like! A universal chat/voice bar for everything and embedded apps? How do apps evolve when you don’t have to click/tap?

@dittycheria It’s possible to make the chatbots solve even complex requests. But the problem is often companies are too careful and slow and don’t give the bots enough tools/APIs to take more actions. Most companies don’t even have the APIs available.

I have yet to use an AI assisted or automated customer service chatbot that works. They all fall apart beyond basic requests, and god forbid you have a corner case.

ChatGPT basically wrote code to create a deck and overcomplicated it quite a bit. The slide quality was pretty good though. I went with Manus and it was also nice that I could transfer it over to Google Slides and continue editing there. Very cool product overall!

Gamma's deck was "pretty" but didn't quite have the enterprise feel to it. But it was fast and super easy to use. Manus gave a really solid enterprise grade deck and was easy to make edits through prompting but didn't feel like I had much creative control.

I asked Gamma, ChatGPT and Manus to prepare a sales deck for me and gave all of them the same main prompt and a deck outline. Gamma took 10s Manus took about a minute ChatGPT took a staggering 52 mins

@bcherny What’s the benefit of doing this for you?

2/ I also run 5-10 Claudes on claude.ai/code, in parallel with my local Claudes. As I code in my terminal, I will often hand off local sessions to web (using &), or manually kick off sessions in Chrome, and sometimes I will --teleport back and forth. I also start a few sessions from my phone (from the Claude iOS app) every morning and throughout the day, and check in on them later.

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.

This was a surprisingly good read and makes so much sense when you look at the world from this lens. Gave me lots to think about and how the world will likely shape up in the next few years if some of these trends continue.

@bhalligan @dittycheria

Who are some non-obvious CEO guests I should have on my pod?

As 2025 winds down, sharing some practical lessons I learned after a year of building AI agents in production: linkedin.com/pulse/lessons-…

I’ve gone from Google search —> ChatGPT —> Gemini. GPT 5.1 feels just a bit too verbose and over eager for my taste, especially for simpler search like queries

My X algo has gone from educational, funny and entertaining to mostly just rage inducing

Also “Here’s the kicker”

Early adopters have already integrated AI into their current use cases and will want to carefully evaluate new versions before switching over. The switching cost needs to feel like it’s worth it. My startup for example has deeply integrated O3 and we understand it somewhat intuitively at this point. When we tested GPT-5 on our internal evals, it performed significantly worse, probably because it needs its own set of prompt retuning which we don’t have time for if it’s only marginally better than O3.

As model wars heat up, I suspect the first mover advantage is fading away and AI labs will want to wait longer to launch the next version. Expectations are higher than ever. Perception around scaling laws hitting a wall can hurt their ability to attract more capital.

One of the best use cases for really demonstrating the power of AI to the masses is if it handles something like filing taxes since most adults need it. I don’t think models are quite there yet but this type of use case could be a good milestone to measure progress towards AGI.

People still mostly use chatgpt as a Google replacement. New devices and more ways to integrate AI into your everyday workflows will be necessary to bridge the overhang between models and applications. There’s still so much build in the app and devices layer!

Even as models get significantly better in raw intelligence and get closer to AGI, most people will not really feel the intelligence leaps because 1) most people don’t need advanced intelligence for every day search queries 2) chat is a very limiting interface overall.