Abhinav Anand

Null & Void, Ex Ex-Tech leader at Meta, Curefit, Amazon; 2nd time founder; IIT KGP CS Grad; All about AI and Tech;

[Abhinav Anand @ab_anand10]

@_sam22 @Uber_India @UberIN_Support @PoliceBangalore @UberIN_Support can you help with this please

[Abhinav Anand @ab_anand10]

🏏The reason we all love IPL so much! 🏏

The thrillers like LSG vs DC is the reason I get glued to the television during the IPL season. A complete no-body comes and grabs the victory from opposition’s jaw and becomes an overnight talk of the town.

Ashutosh Sharma came in when DC were reeling at 65/5 in 6.4 overs with a mammoth chase of 210 ahead of him. He started slowly and was happy playing second fiddle while Stubbs and Vipraj went hard on the opposition. When Vipraj and Ashutosh were at the crease, it seemed DC might cruise to an easy victory, but IPL is never short of drama and soon, Ashutosh was left with Vipraj’s, Starc’s, Kuldeep’s wickets tumbling around him, while he took over the power hitting to himself.

Soon it was 6 required of last over with 1 wicket in hand and Mohit Sharma is the last man facing the first ball of the over. Ball one goes up to the 3rd umpire because Mohit was most likely caught plump in front of the stump. But as luck had in fate, it was Ashutosh’s day and Mohit Sharma was saved by a whisker.

Mohit somehow stole a single and it was Ashutosh back at the crease and the first ball he strikes the cleanest six you will ever see in cricket.

That six was the reason we all watch IPL and maybe even cricket.

The confidence in his eyes, the eagerness to prove himself, the calmness of a sadhu is what every human strives for and Ashutosh was the symbol of the same at that very moment. These are the moments we all wait for the entire duration of IPL’s three and half hour long encounters.

His confidence is evident in how comfortable he was letting Mohit take five balls to score a single, leaving just the final delivery to win the match—as Mohit Sharma later revealed in the post-match interview.

Ashutosh is a player to watch out for this IPL season. He is hungry and raring to go. Not sure why Punjab didn’t retain him, but I feel he has found a better home at DC under Kevin Pietersen. Really excited to see what more these guys will be cooking up this IPL season.

#IPL2025 #DCvLSG #AshutoshSharma

[Abhinav Anand @ab_anand10]

Sharma ji ka ladka 👏👏👏

#DCvLSG

[Abhinav Anand @ab_anand10]

With all this load of crap about AI, why does my Google Home & Alexa still suck at playing "Na Na Na Re By Daler Mahendi"

Isse simple lyrics kya hi honga 😛

[Huku AI @HukuLabs]

1/ A deep dive into why reasoning models (O1, DeepSeek) represent a fundamental shift in AI interaction. From my experience working with these models, here's what you need to know about maximizing their potential 🧵

[Abhinav Anand @ab_anand10]

@parasharmrigesh The only way to survive this onslaught of LLM coder agents is for all us humans to write crappy documentation and crappy API schemas ... we are all in this together .. let's fight the force 😀😀😀

[Abhinav Anand @ab_anand10]

This is called the `Power of Distribution`. Putting the AI Icon in the search bar has made me accidentally click Meta AI so many times, even when I just wanted to search. Master stroke Zuck :)

[Abhinav Anand @ab_anand10]

@parasharmrigesh How badly designed is the world's most important OS that it can be brought down by a small single line commit.

[Abhinav Anand @ab_anand10]

not sure where is the hate coming from. Picking up a number of 35mn$ and saying that’s absurd. Then we can all come up with umpteen examples of where billions of dollars have gone into much worse things.

I have used langchain and langgraph and found it to be really useful. All software stack is an abstraction be it css to tailwind or html to react. It’s just another way to organise your code. Either you write all the helper utils yourself or use libraries like Lang chain to not worry about the code orchestration.

It should be a developer’s choice to decide which library meets his or her requirements based on his research and experimentations.

[Abhinav Anand @ab_anand10]

@parasharmrigesh 😂😂

[Abhinav Anand @ab_anand10]

@paulg who are these people who care whether Sam was fired or not. He ran YC, world’s largest startup accelerator and now is building AGI. Let the rest of the story remain a mystery. Makes for a good spicy suspense in the “Book on Sam Altman” 😂

[Abhinav Anand @ab_anand10]

@parasharmrigesh like oil fields in Middle East 😂😂

[Abhinav Anand @ab_anand10]

@parasharmrigesh so eerily close 😀😀

[Abhinav Anand @ab_anand10]

"I am gonna eat you all" says OpenAI

OpenAI just made a massive entry into consumer game with the launch of Mac App last week. We all focused on the ‘o’ in the GPT-4o, but OpenAI just changed the whole landscape. A mac app, that sees it all, understands it all and maybe in future acts on all, is the mother lord of the new generation of LLM based OS.

Till the time Apple’s Siri comes out of it’s slumber and realises what is expected of a voice based assistant, it gives good enough time for OpenAI to run wild and build a solid consumer presence with it’s conversational agent offerings across all platforms.

Everybody thought OpenAI primary strategy is to be a developer friendly AI cloud service provider. They were first and most powerful LLM provider in the market. OpenAI’s entry into consumer game, just highlights the fact that their is gonna be very low value in just building world-class models (the reason https://t.co/LYAvJ3oC46 had to be absorbed into Microsoft rather than competing in this fierce low-margin model serving business). Maybe a few top niche model providers like ElevenLabs have a chance, but most of it is gonna get commoditised very soon and offered at dirt cheap price on AWS.

OpenAI’s mac-app and voice-based human interaction suddenly puts them on a head-on collision path with Google and Apple, maybe even their pals at Microsoft should now get a bit worried.

Be warned and cautious, OpenAI is not just coming for Google and Apple, they are coming after all language translation apps be it DuoLingo, they are coming after all edTech products, they are gonna steamroll all the chatgpt wrappers, they are gonna kill every character building AI startups, almost anything related to language is in the firing line.

Nobody is gonna pay for multiple AI service providers, it’s gonna be winner takes all market, similar to Facebook in social commerce or Apple in phone commerce. Very few AI services would be alive when the dust settles. SaaS, Business process automation and other niche use-cases like generative AI editing tools are a different category but consumer facing generic image generation, language related AI use-cases are gonna soon fall under a single OpenAI banner

So, beware and stay as far as possible yet navigating around OpenAI’s ring of fire, because that’s where the new generation of Software is gonna be rewritten so you just cannot avoid OpenAI.

We have an interesting coverage of all the interesting features of OpenAI GPT-4o model on our website. Do take a quick read.

#openai #gpt4o

https://t.co/cLMzVGeop6

[Mrigesh @parasharmrigesh]

GPT-4 marks a significant transformation in human-computer interaction, seamlessly integrating AI into our daily lives. AI has become our coworker, companion, and a constant presence, drawing us nearer to the world envisioned in "Her." Developing a deep connection with AI now seems like an inevitable progression.

[Mrigesh @parasharmrigesh]

I agree with you, but I also disagree with you.

Why does an AI wrapper work? An empty textbox can feel overwhelming, and most people don't really know how to make the best use of it. What an AI wrapper does is cater to the needs of a specific group. It's the small details that make the product truly remarkable.

With AI wrappers, you might not build a VC-scale company, but you can definitely build a profitable small company around this.

I completely agree with the human-computer interaction and AI agents part. Most AI wrappers can be templatized, and 80-90% of your work can be achieved by a group of agents.

However, the outcompeted part only comes into play when you want to build a billion-dollar company. Most users don't need advanced cameras, but a niche of photography enthusiasts still prefer DSLRs over phone cameras.

[Abhinav Anand @ab_anand10]

A desktop app that can see anything running in the background with a voice based interaction just feels so much like a real co-pilot. Soon when it can start taking OS level actions on our behalf, voice would truly become our go to way to truly interact with computing forever and ever !!
Open a webpage and boom !!

#openai #gpt4o #springupdate

[Abhinav Anand @ab_anand10]

@garrytan Zapier might be doing the plumbing but there is a long way to build consumer-grade knowledge workdesk that can allow users to interact with multiple agents. Knowledge work is gonna get super complex. Exciting times ahead !!

[Abhinav Anand @ab_anand10]

The Two Flavours of B2C SaaS Products: B2C2B vs B2B2C

When you consider building a SaaS product for consumers, be clear to whom are you going to sell first. Are you going to sell to the business first or the individual first ?

In B2C2B, the focus is usually on acquiring individual users first, who then promote the product within their organisation. A great example of this is Notion, which often enters organisations through individual users before being adopted company-wide. Prosumer is another word used to categorise this set of “Professional Consumers” who bring products into their organisation.

Even open source developer projects are B2C2B because they are first used by an individual developer. You need to package, document and provide high quality tutorials for them or else developers will never be able to use your product. Slowly you spawn off a commercial offering based on the open source project similar to Kafka, Langchain, even OpenAI is developer first than business latter.

No code tools are also great example of B2C2B, how they slowly get adopted by founders and very quickly become the lifeline of the companies. Most of the future AI applications would end up as No-Code Tools providing you agents and ways to connect agents to your toolkit.

On the other hand, B2B2C SaaS involves selling a product to a business, but it's ultimately used by the end consumer. A great example of this would be Microsoft Office Suite, which sells its services to businesses, but the product is ultimately used by the consumers who are employees at these companies

What you should look out for while building B2C SaaS product ?

The primary difference between the two is not just the GTM but also in the product-development life-cycle.

An individual might care about a better user experience (UX), business might put more focus on security and compliance.

An individual might be willing to discard his current toolkit and completely disrupt their current workflows, business would care about integrating with their existing toolkit and a product that fits into their existing business workflows as a piece of a puzzle.

An individual might be more daring to trust you with their data but not businesses.

Both ways of building might lead to a very similar "End Product", but the difference happens in the feature prioritisation and GTM strategies at the early stages of the product life cycle.

Will you launch on Product Hunt or sell to CIOs ?
Will you build compliance first or WhatsApp bot ?

It’s very critical to identify your product’s gtm strategy and prioritise accordingly. Most of the startups die due to lack of prioritisation. If you don’t have your GTM sorted, probably good time to start thinking about it !!

#entrepreneurship #gtm #startups #ai #aiagents #nocode