Govind Chandrasekhar

Its so much easier to talk about stuff than write it up well. Not sure if a followup article on the India/OpenAI topic will ever happen - but a lot of it got covered chatting with @govind201 and @rohith_salim on the Weekend Entrepreneur podcast. Podcast Link in thread.

@airmauritius has got to be the biggest scam around. Flight cancelled just 24 hours before the trip. No word on a refund for 2.5 months. Cancelled hotel expenses are your problem any way. Easiest way to run a business -> take a customer's money, decline service & disappear.

@WhatsApp 2/WhatsApp doesn't have any rapid escalation mechanisms for such time-sensitive situations. We tried mass blocking her account & resolved it. But WA is such critical digital infra in places like India - it's scary that UX flow decisions can have a huge impact on people's lives

1/My aunt’s @WhatsApp was hacked through social engineering. Before we could log back in, the hackers initiated many failed login attempts, leading to WhatsApp blocking logins for 10 hours. This left us, the legitimate SIM owners, locked out while the hackers ran their scams!

The Most Important Question to Answer Professionally For Yourself (and Others): ~What do you REALLY care about~???

@ParikPatelCFA @apoorv03 "what did WAU/MAU look like at similar stage & scale early in their lives" This question is somewhat answered by another graph in the article. Any thoughts on this?

I think your charts may support the thesis that ChatGPT specifically lacks engagement required to win in same way as other consumer products, but there are lots of problems with this when you broaden this to AI apps. You need to first define what winning consumer AI means. This is also same thesis everyone articulated two years ago and the charts you are using do not really articulate point - you are comping already established & densely used product engagement with more nascent products with building habit formation - ie, of course new products don’t have the metrics reflected by very successful consumer products today. Your chart with adoption vs engagement kind of proves this point. You need to find base rate that is appropriate comp in early years for this analysis to be useful: eg for the consumer products deeply embedded in people’s lives today, what did WAU/MAU look like at similar stage & scale early in their lives? Same with retention. That data is prob not available / hard to find. You then jump to conclusion that because Meta and Google are already entrenched in people’s lives, they are positioned to win. This same argument works for Apple and Microsoft (iOS and windows are arguably even more entrenched). People don’t go to FB or IG or Tiktok for search which is pretty big habit formation hurdle. In theory Google is primed to win because they already have consumer habit formation and entrenchment but they haven’t executed which is why we are all having this debate. TLDR is nobody knows anything and new players prob have higher probability of winning than implied.

Why Meta & Google may win Consumer AI, before ChatGPT does! A 🧵 1/ ChatGPT burst onto the scene as the killer AI app, setting off alarm bells across the tech world. But while its run has been impressive, usage metrics suggest that Meta/Google/Tiktok have much better engagement

There are three ways to back or build within the new thing: foundation, infrastructure, and applications. The most interesting questions and best businesses are always answers to “what becomes possible in a world where XXX is better, cheaper, faster” But we can see the same race to middleware play out from crypto, to AI, to fintech, to robotics, etc. Everyone wants to monetize early “activity” without a specific point of view.

@angadsg Not this much faster. I have a batch job that runs roughly 3 processes regularly. This is a few magnitudes faster. I might be wrong.

@govind201 It gets faster on the weekend in non US Timezone due to lower traffic. Have stats?

Is it just me, or has GPT-4 Turbo gotten 3x faster over the weekend?

If you're not in on it, almost no startup that says it's "crushing it" on LinkedIn/X feels that way. It's the pressure to portray success, which is necessary for marketing/fundraising/hiring. In private, ~everyone admits to stumbling along, just some more gracefully than others.

@trammyyy Thanks Trammy!

In 2013, I raised from YCombinator to build an Ecommerce AI startup. A decade on, I did it again! At Branch-AI.com, we help unlock retailers & brands state-of-the-art search, without needing armies of relevance engineers, using LLMs as orchestrators. Demo on website!

@pravinmiishra @ycombinator @ishanagr We use LLMs to generate training data for embeddings, rankers & query regeneration. First, LLMs are quite resilient to out-of-domain needs. Second, we fine-tune them on multiple auxiliary tasks constructed from behavioral data (e.g. reviews, query-product pairs, 0-result queries)

@ycombinator @govind201 @ishanagr How does your system adapt to dynamic product offerings and shifting customer preferences?

YC S23's Branch AI (branch-ai.com) helps ecommerce companies unlock state-of-the-art search, instantly. Their AI learns product catalogs and tunes itself to maximize relevance and conversions. ycombinator.com/launches/KU0-b… Congrats on the launch @govind201 & @ishanagr!

@sachin_siby Thanks Sachin!

Congrats on the launch @govind201

@TomAnswerAi @ycombinator @ishanagr Our focus has been on the end retailer, but it's APIs under the hood so the system is flexible. My DMs are open if you'd like to dig in?

@ycombinator @govind201 @ishanagr Open to third parties to enhance our Shopify chat? Or just focused on end retailer @govind201

"We do these things not because they are easy, but because we thought they were going to be easy" is a surprisingly profound quote. When I ask people who've pulled off remarkable things, it's interesting how many confirm that they wouldn't have started if they'd know how long and difficult the road would be.

@paulg Would your advice hold for SaaS/B2B businesses, for whom the customer is usually a US company?

I feel a bit out of touch with current fashions, because a couple days ago an Indian founder in the current batch asked if he should base the company in SF or India, and I said the Indian startup community seemed to be thriving and it would be fine if he wanted to go back.

The way we search is changing - from keywords to intent. We see this in an increase in average search query length and the number of unique queries. This has 2nd order effects on buyer journeys, user analytics, personalization, SEO/SEM & AI investments. blog.branch-ai.com/second-order-e…

@sachin_siby Not sure I'd put it that way personally. Do you think there's an inherent difference between different games? Or is that your preference overlaid?

@govind201 In short — are you optimizing for dopamine or serotonin

6 games one can play in life — mission, mastery, autonomy, stability, influence & money. Discontent stems from playing the wrong games, or by playing too many at a time. @govind201/life-games-2c668db07287" target="_blank" rel="nofollow noopener">medium.com/@govind201/lif…