Harsheet Singh

Trying to make AI agents less demo-ish and more useful

Building · IITB

Notes from building, breaking, and overthinking product

[Harsheet Singh @harsheet46455]

@markproduct Professional Representation

[Harsheet Singh @harsheet46455]

A friend is dealing with this after paying in full. @royalenfield, please help resolve it quickly and ensure the dealership accounts for every payment. Sharing for visibility.

[udasi @onesquarepeg]

@royalenfield
Worst experience placing an order at your authorised dealership in Bellandur - Teknik Motorcycles.

I’ve paid the full amount in advance, but they’re not giving me my order and are asking me to pay Insurance, RTO and accessories amount twice.

[Harsheet Singh @harsheet46455]

Lean can verify whether the encoded proof follows.
It cannot verify where the decisive idea came from.
Agentic research needs correctness and provenance. A verified answer is not automatically a well-attributed answer.

[OpenAI @OpenAI]

We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.

The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.

The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.

[Harsheet Singh @harsheet46455]

@BehereBaba Waiting for a Micromax Made in India foldable at ~95% discount with 200% increment in features

[3one4 Capital @3one4Capital]

https://t.co/MiN7jcnsec

[Harsheet Singh @harsheet46455]

@anujrathi @ETNOWlive Incredible to see the vision coming to life. Excited for everything we’re building ahead! 🚀

[Harsheet Singh @harsheet46455]

Building the future where every great professional has an AI rep working for them.
Exciting to see the vision getting out there. Just getting started 🚀

[Anuj Rathi @anujrathi]

We're building https://t.co/oeXGgZxEdq - Professional AI reps for top professionals! Check out what the future looks like!

Thanks, @ETNOWlive for covering our vision for the future of work.

[Harsheet Singh @harsheet46455]

Most professional platforms index credentials. We’re building Profound to understand the judgment, context, and ambition behind the work.
Proud to be part of the founding team building this in India, for the world. And today, I get my own Profound Rep too🚀

[Anuj Rathi @anujrathi]

Meet https://t.co/WqFto9VLas. Every professional gets their own AI rep, one that learns your work and goals in a 30 minute conversation and carries it into the roles, rooms and people worth your time.

Reserve your Profound AI Rep, and tell me how your first conversation went!

[Harsheet Singh @harsheet46455]

@anujrathi Months of hard product questions, rapid prototypes, and plenty of rebuilding went into this launch. Proud to have helped engineer Profound from the ground up alongside this team. Now comes the best part: seeing people meet their Rep for the first time🚀

[Anuj Rathi @anujrathi]

Meet https://t.co/WqFto9VLas. Every professional gets their own AI rep, one that learns your work and goals in a 30 minute conversation and carries it into the roles, rooms and people worth your time.

Reserve your Profound AI Rep, and tell me how your first conversation went!

[Harsheet Singh @harsheet46455]

Launching a membership program just because a competitor has one is one of the most common traps in consumer tech. You usually just end up bleeding margin without building any real retention loop. Anuj breaks down the actual mechanics of doing this right below.

[Anuj Rathi @anujrathi]

The hardest moment in a membership program arrives a few quarters in, when the P&L starts looking bad and the pressure to pull back or dilute the program builds.

Whether the leader stays put depends on the preparation: benefits packed to be desirable, predictable, and profitable over the long term, and a willingness to run a new operating system for the company rather than a "growth bet."

We talked about the right shape of business for a membership program, the preparation it demands, and how to build one that lasts (also Swiggy One), with @rchowdhri on Interstellar by Stellaris.

https://t.co/j8bir5ZwaH

[Harsheet Singh @harsheet46455]

@aryanXmahajan OS

[Harsheet Singh @harsheet46455]

@BehereBaba The next question might be: "Is it worth changing the workflow for?"

[Harsheet Singh @harsheet46455]

300,000 Copilot seats is impressive.

What's more interesting is that the conversation has shifted from:

"Which AI tool should we buy?"

to

"Which workflows should we redesign?"

That's when a technology starts moving from software to infrastructure.

[India Plus @india_plus_]

🚨 TCS, Infosys and Wipro have collectively acquired more than 300,000 Microsoft Copilot licences in just 6 months,

Marking a major push towards AI adoption in India’s IT sector.

follow @india_plus_

[Harsheet Singh @harsheet46455]

Everyone is racing to build autonomous agents.

Users seem more interested in retaining control.

Gartner found only 11% of consumers are willing to let AI make purchase decisions for them.

Maybe the future belongs to copilots before agents.

[Harsheet Singh @harsheet46455]

Agentic AI has a supply boom, not a demand boom yet.

Everyone is shipping agents.
Very few are becoming habits.

The gap is not autonomy.
The gap is repeat value, agents useful enough that people come back the next day.

[Jen Zhu @jenzhuscott]

Massive output uptick due to agentic AI. Complete flat adoption.