Pinned Tweet
Another AI tool compiler? Why bother? 🤔
Glad you asked! Let's talk 'Google For AI' 👋
#buildinpublic #chatgpt
Such a wonderful start to Insurjo'24 by @TheProductfolks 🙌
Thank you for hosting such an insightful session by Rishi Bhargava💎💲
I can't wait to finish my Pizza Treat and start drafting the session notes 😤
4
63
Adityaraj Ray retweeted
Startups vs Large Scale Companies. 🏢
Can they be anymore different?
With different goals and product infrastructure, they're at opposing ends of the spectrum.
So, how does a Product Manager prepare for them both? 🤔
⏰ In less than 5 minutes, Rishi Bhargava, Co-Founder, @descopeinc and ex-VP at Palo Alto Networks will be joining us to shed light on this nuanced topic and delving into all these intricacies.
📽️Join us here - rebrand.ly/Insurjosession_Ri…
We’re ready to live tweet! ✍🏻 Let's go!
Adityaraj Ray retweeted
Hot take:
Building cool shit on weekends is more fulfilling than waiting to shut your laptop on Friday, partying all weekend, waking up hungover, and hating Monday.
Adityaraj Ray retweeted
An aspiring PM asked me: "How can you create products if 'everything' we can think of is already invented?"
My answer: The problem with that question is that it starts by thinking about products, not problems to solve, much less about users.
People → Pain Points → Solutions
Adityaraj Ray retweeted
We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we announced at Microsoft Ignite, and in continuing to support our customers and partners. We look forward to getting to know Emmett Shear and OAI's new leadership team and working with them. And we’re extremely excited to share the news that Sam Altman and Greg Brockman, together with colleagues, will be joining Microsoft to lead a new advanced AI research team. We look forward to moving quickly to provide them with the resources needed for their success.
This does not feel like the final we deserved!
Can we do a re-run at Wankhede?
With MSD on the field 😭
#INDvsAUS
1
137
I've curated a list of 9000+ GPTs.
The list has GPT Names, Description & Links.
Want it?
It's FREE for 24 hours!
To get it:
👉 Retweet
👉Like & Reply "GPT"
👉Follow (for DM)
#giveaway
1
1
55
Adityaraj Ray retweeted
Rumor: ChatGPT is getting a memory.
This is a HUGE deal. It is on the way to becoming a copilot for your mind.
This feature takes all of the power of Custom Instructions and makes it 10x more valuable.
- never have to explain yourself twice—ChatGPT will remember who you are, and all of the important people in your life
- use ChatGPT as a repository for notes, ideas, quotes and more.
- it will resurface all of these in new chats where relevant. A true second brain.
The best tool for thought is tools that think 🚀
Cannot wait for this to drop.
Not a very comfortable win for India
Lot's of extras
Lot's of misfields
Lot's of no-man's land
Lot's of catch-drops & miss-fields
However, India chose to win. 🫡
#IndiaVsNewZealand
62
4.9Cr people watching the match,
only for India to take Mitchell's wicket.
Let's not allow him to do a @G_Maxi32
#IndiaVsNewZealand
1
88
Who is playing? @apdhillxn or Siraj?
No wonder K. Maharaj overthrew AP - WR1
All we need Siraj to do, is take out Mitchell
Eh munde pagal ne saare,
Gallan vaddiyan vadde laare
#IndiaVsNewZealand #INDvsNZ
4
821
With Williamson down, the fall of the house begins. If Shaami is able to topple Mitchell, we could potentially see a >100 run victory.
Shaami potential will take 5 wickets.
#IndiaVsNewZealand #INDvsNZ
1
251
So many potential catches in no man's land. So many potential wickets.
So many catch drops.
So many extras.
WTF is wrong with India today?
All they need is 1 wicket, and
There topples the House of Cards.
#INDvsNZ #IndiaVsNewZealand
Adityaraj Ray retweeted
Thinking a lot about what the implications for startups are after OpenAI day.
Here are a list of my thoughts:
The dunbar number for chat is low—probably 1-2. In other words, consumers will only actively remember to go back to 1-2 chatbots (maybe one for work, one for personal.) The chatbot that wins is the one that produces the smartest responses most consistently across queries. OpenAI will in because it is in the best position to improve queries quickly because…
It owns the foundation model and ChatGPT. ChatGPT is the biggest consumer AI chat product, so it produces an insane volume of data to directly improve the underlying model. OpenAI integrates a density of cash, talent, foundation model tech, and consumer users—which are the crucial ingredients for improving model quality the most. High ChatGPT usage creates more demand as model responses improve.
Developers should not try to build general purpose chatbots to compete directly with OpenAI.
Developers should not try to build niche chatbots, unless they will be used at least a few times a day by a particular subset of people or they are integrated into some other product experience.
No one will remember to use something they only have a question for 1-2 times a week or a month. AI chat works when integrated into existing products, but only 1-2 will become a destination.
In this way, the chatbot category will look like search—there are search boxes on every website, but there’s only one Google
Developers should not be building on the knowledge orchestration or the evals layer of infrastructure for OpenAI.
OpenAI needs to improve model response quality for developers.
The best way to get better quality responses is 1) data (from ChatGPT) 2) money to train large models 3) researchers to deliver breakthroughs and 4) knowledge orchestration 5) evals
OpenAI as a strategic priority must do great knowledge orchestration (like integrating retrieval) and evals (which it’s already doing, and improved in its interface after Dev Day.) These will continue to get more advanced, and they will beat out modular competitors because OpenAI controls the entire stack and will have the ability to tune them to perform well.
There is still room for infrastructure at this layer of the stack for developers who are focused on open source alternatives, but open source, modular players will only be relevant as the S curve of improvement for model performance starts to level out. Once that happens, there will be a focus on price and flexibility rather than pure performance, and modular players will be in a reasonable position.
There are risks to OpenAI’s power, though. Specifically, because their org size is going to expand, and they need to consider geopolitics etc. as they decide what to release. In AI, paying attention to safety is a risk to adoption and makes companies vulnerable (see DALLE vs. Stable Diffusion / Midjourney as one example). Companies with less to lose will be able to take more risks, and that will be a way for them to improve model performance even without being on the technological bleeding edge. In other words, technology will soon not be the core limiter of model response quality, risk tolerance will.
OpenAI may find clever ways to offload this risk, custom GPTs is one example that might let them avoid some copyright issues. But it’s inevitable that they’ll get more sensitive as they scale.
There are significant opportunities to build inside of the OpenAI ecosystem. If ChatGPT is iOS we don’t yet know what the App Store is. OpenAI has tried plugins, and their latest attempt is GPTs. Imo, until they introduce a ChatGPT experience that lets users quickly switch back and forth between vanilla ChatGPT and a custom GPT in the context of the same chat (and maybe do that automatically depending on the message), adoption will be low.
















