amitk

🚀 Founder @ EarnlyAI ( | AI-native commerce for Shopify | Upshot • Subnexa • NudgeX | Turning customer intent → action → outcome |

[amitk @kl_amit]

Spot on, Gabriel. This is exactly where intent-to-action breaks. “Navy under 20k” has to resolve to the navy variant: its price, its stock, and its photo. That is how we are building Upshot to work, and the missing colour photos are the other half of the problem that is getting solved with Acuarela.

[amitk @kl_amit]

E-commerce search is due for a rethink. 🛍️

Shoppers don’t come to search. They come to achieve something.

Think about it: someone shopping for a winter wedding doesn’t just want a dress. They need an outfit for themselves, a gift for their brother, and everything within a ₹20,000 budget.

That’s not a search query. It’s a combination of intent, preferences, constraints, and a desired outcome.

Yet, the traditional journey makes shoppers search, filter, compare, search again, and finally add to cart. Too much effort between knowing what you want and actually getting it.

The future of e-commerce isn’t just better search. It’s moving from intent to action.

At EarnlyAI, we’re building AI-native commerce experiences that understand what shoppers mean, work with live product catalogs, inventory, variants, and pricing, and help turn that intent into real outcomes.

Type it. Say it. Show it.

Less friction. Smarter discovery. More completed outcomes. 🚀

That’s the shift we’re building with Upshot for Shopify. https://t.co/ix0C02fFnd

#AI #Ecommerce #Shopify #AgenticAI

[amitk @kl_amit]

We have tested Upshot against a research-driven approach to evaluating AI shopping agents.

At EarnlyAI, we believe building an AI shopping agent is not just about making it respond intelligently. It must understand shopper intent and help customers make the right purchase decisions reliably.

That is why we evaluated Upshot using insights from ShopGym, a research framework for realistic simulation and scalable benchmarking of e-commerce web agents.

The paper addresses an important challenge: testing on live stores offers realism, but results can be difficult to reproduce. Simplified test environments are easier to control, but may miss the complexity of real shopping experiences.

- ShopGym tackles this through two components:
- ShopArena: Creates realistic, repeatable sandbox storefronts.
- ShopGuru: Generates shopping tasks grounded in store structures, product catalogs and policies.

The research evaluates 224 tasks across six sandbox shops and reports a positive correlation between agent performance on synthetic shops and corresponding live storefronts.

For Upshot, this research provides a valuable lens for evaluating important aspects of the shopping journey:

- Understanding shopper intent and finding relevant products.
- Applying budget constraints and product filters.
- Answering questions using product information and store policies.
- Handling multi-step shopping journeys rather than isolated prompts.
- Identifying failure scenarios and improving consistency through repeatable evaluation.

Our objective is simple: move beyond impressive demos and build an AI shopping experience that is useful, reliable and measurable.

At EarnlyAI, we believe AI quality is not just about what an agent can do. It is about how consistently it can do it.

Explore Upshot: https://t.co/otDcCpOgG3
Available in Shopify App Store: https://t.co/HZUfPvxqbE

Research paper: https://t.co/IK0fFo2D2V

#EarnlyAI #Upshot #Shopify #AgenticAI #Ecommerce #AIQuality

[amitk @kl_amit]

We(Upshot) are working on a new way to shop on Upshot, search by how things look.

Your customers search in pictures. Your store searches in words. We're working on fixing that.

Shoppers don't type your product titles. They type what they picture.

"Oversized tee with a red graphic on the back"

We are teaching Upshot to find it in your photos, not just your titles. 👀

🚀 Try Upshot on Shopify: https://t.co/N5ahwKBlnA
🌐 Discover EarnlyAI: https://t.co/k2n1MedqGD

[amitk @kl_amit]

@peloranapp Thanks. That’s the idea. Upshot already pauses on a full cart and offers help with the shopper’s actual items, and it also follows up by e-mail if they leave. The difference is the help is specific to what’s in the cart, not a generic “need help?” pop-up.

[amitk @kl_amit]

Most ecommerce stores don’t lose customers because of bad products. They lose them because shoppers get stuck. 🛒

So we built Upshot by EarnlyAI to step in at the right moment. A gentle nudge. Answers about their cart. Help before checkout.

Not another popup screaming “BUY NOW.” Just AI that actually helps shoppers buy. 🤝

That’s how we think ecommerce should evolve.

🚀 Try Upshot on Shopify: https://t.co/N5ahwKBlnA

🌐 Discover EarnlyAI: https://t.co/k2n1MedqGD

#Shopify #Ecommerce #AI

[amitk @kl_amit]

Every code changes and feature from https://t.co/EQcACRo7VT goes through following sub-agents for ensuring quality and secuity compliance.

We are an AI Native Organization and our agents are there always review and fixing in an unbiassed manner with full auditability.

[amitk @kl_amit]

One week of Upshot being live on the Shopify App Store.

And we are learning a lot. We built Upshot around a simple idea:

An AI shopping assistant should not just answer. It should help the shopper buy.
- Find the right product.
-Check stock.
-Respect the budget.
-Compare options.
-Build the cart.
-Checkout.

And do it using the store’s live catalog, not AI guesses.
Chat, voice, photo → product → cart → checkout.

We are still early, but the first week has been exciting.
If you run a Shopify store, I would love to know:

What is the #1 question your shoppers ask that you wish AI could handle better?

Try Upshot → Shopify App Store https://t.co/rtU28I6n9X

Or try it yourself → https://t.co/LuCVDp1w7r

Built by EarnlyAI (https://t.co/Zp17wn8YZy)

[amitk @kl_amit]

@CharuMitraDubey We are building Intent based Conversation Agent https://t.co/dhGBEZhs4u and AI Native Subscription Management https://t.co/SoNCh5uvFA

[amitk @kl_amit]

tagging @ShopifyDevs @Shopify

[amitk @kl_amit]

🐯 What if your Shopify store could understand what customers want — not just what they search for?

That’s what we built Upshot for. Now live on Shopify. 🚀
Intent → Action → Outcome 🎯

👉 Explore Upshot
#Shopify #CommerceAI #AgenticAI https://t.co/8zSIYO8fD5

[amitk @kl_amit]

tagging @Avishek94652808

[amitk @kl_amit]

As I am progressing more with AI Native Subscription Management application coupled with Shopify, I am realising that shopify store owners are spending more time on tech work with defining Subscription & managing them.

\What if, there are agents you define who does all the work for you and come up with Proposals along with ROI and you review, edit and approve? They also learn along side from your feedback and will become more accurate as time goes.

This is what we are building with Subnexa https://t.co/MiVmuGfXib and it is really becoming an interesting one. We will wrapup this soon and will be available to shopify store owneres in limited Availability.

Let store owners focus in Product management and Store Management while your owners does the heavy lifting for you to manage the subscription plans and monitor them...

[amitk @kl_amit]

🔄 Subscriptions shouldn’t just renew. They should think, adapt & act.

With my background as a CTO building Cloud & Agentic AI platforms, we’re taking a different approach with Subnexa — AI-native and agentic by design.

#Shopify #earnlyai-subnexa https://t.co/ZIQYH7feYh

[amitk @kl_amit]

🔄 One thing I’ve been thinking about while building at EarnlyAI — why should subscription platforms only manage subscriptions?

They already have signals around renewals, failed payments, skips, cancellations and customer behaviour. An AI-native platform should be able to understand those signals and do something useful with them.
Having spent years as a CTO building cloud platforms and Agentic AI systems, I wanted to approach subscription management differently.
That’s the thinking behind Subnexa.
Not adding an AI chatbot on top of traditional subscription software — but making the platform AI-native and agentic by design.
🤖 Observe what’s happening
🧠 Understand risk & intent
🎯 Decide the next best action
⚡ Act through the right workflow
🔁 Learn from the outcome
All while keeping the merchant in control.
Manage → Understand → Decide → Act → Learn
Still early, but I believe subscription management is a very interesting problem for Agentic AI to solve.
Would love to hear what Shopify merchants think.
👉 Explore Subnexa
#Shopify #AgenticAI #SubscriptionCommerce #AINative

[amitk @kl_amit]

🔄 Subscriptions shouldn’t just renew. They should think, adapt & act.

With my background as a CTO building Cloud & Agentic AI platforms, we’re taking a different approach with Subnexa — AI-native and agentic by design.

#Shopify #earnlyai-subnexa https://t.co/ZIQYH7feYh

[Andrej Karpathy @karpathy]

We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does anything at all.

I also like this kind of examples because no one in their right mind would ever spend the time to write something this custom but LLMs have all the stamina and patience in the world, so it's an example where we go from "no one would ever do this" to "sure, why not, it's ~free". There might be a lot more. But I'm excited about creating hyper custom worlds that you can imagine dropping players into, e.g. here to participate in the LoTR story as a spectator NPC, or one of the characters, or etc. Something like an ephemeral GTA of X on demand.

Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Here, Opus 5 had to very slowly and painstakingly take screenshots at different points, and it messed up a few times and created a bunch of jank. An example of raw capability (multimodal, gameplay) that I think is still quite lacking.

[anand mahindra @anandmahindra]

Yeh ladka bahut talented hai.

Kya woh abhi bhi Nainital ki sadkon par kaam kar raha hai?

Main na sirf uski padhai mein madad karna chahta hoon, balki jaadu mein uski dilchaspi ko bhi badhava dena chahta hoon.

Kyon na woh duniya ke behtareen jaadugaron mein se ek bane?

Kya koi mujhe usse aur uske mata-pita se sampark karne mein madad kar sakta hai?

(Video: courtesy @Murti_Nain )

[SaveBellandur(ಬೆಳ್ಳಂದೂರು ಉಳಿಸಿ) @kdevforum]

@DKShivakumar now you got court behind you.Ask every roads in #ITCorridor & #Bengaluru to remove footpath parking,encroachers!
@GBAChiefComm
@comm_blr_south @EASTCITYCORP @Jointcptraffic @osd_cmkarnataka pls implement the order specifically 11,12,13,14.
https://t.co/gN5NXrjWwK

[WINTRACK INC @wintrackinc]

From October 1, 2025, our company will cease import/export activities in India.

For the past 45 days, Chennai Customs officials have relentlessly harassed us.

After exposing their bribery practices twice this year, they retaliated, effectively crippling our operations and destroying our business in India.

We deeply thank everyone who has supported us during these difficult times.

#wintrackinc #imports #exports #chennaicustoms #chennaicustom #airportcustoms #seaportcustoms