Pinned Tweet
We let an AI agent go shopping 101,416 times.
1,044 stores. 25 categories. Every step recorded.
It finished the job about 7 times out of 10. Most failures came at checkout. Sometimes it bought the wrong thing because it never asked a question. Sometimes it gave up when it could have finished.
We wanted to see how agents really shop, and where they break.
Now we're releasing all of it for free. It's called CartTrace-100k.
best team to join
Hiring two founding team engineers in Bangalore.
1. 3D Graphics Engineer. Photoreal 3D walkthroughs in the browser, for real estate.
2. Generalist Engineer. AI agents, datasets and evals. You own projects end to end.
For both, you need to be insanely good at using AI
Reply Here or DM me with the role and I'll send more details 👇
poppy is the front door agentic commerce was missing.
it tells a store who the agent is, who it works for, and what it may do.
i wrote up why that matters.
DM me to get your store agent-readiness score
It’s impossible to ship 3 projects every week without a team.
Super grateful to work alongside
@kushagra_agr @deepanshuxpal @lakshyastwt @devzaidd
Many more updates dropping soon.
Introducing FloorSense.
We've been quietly working on FloorSense with a big mall in Thailand
It's an AI that turns a mall's existing CCTV into a live picture of how people move.
It can count people, see where they walk and stop, and even follow the same person across cameras without knowing who they are, kinda like Google Analytics but for a building.
Still early, but seeing it work on real crowds is pretty crazy.
If you run a mall or any big space, DM me. Would love to try this with more people.
Introducing FloorSense.
We've been quietly working on FloorSense with a big mall in Thailand
It's an AI that turns a mall's existing CCTV into a live picture of how people move.
It can count people, see where they walk and stop, and even follow the same person across cameras without knowing who they are, kinda like Google Analytics but for a building.
Still early, but seeing it work on real crowds is pretty crazy.
If you run a mall or any big space, DM me. Would love to try this with more people.
As personal agents become part of everyday life, merchants have a new frontier for excellent customer service. From answering product questions to supporting purchases, returns, and exchanges, the ambition is to turn a buyer’s intent into an outcome that delights them. Personal Agent Protocol helps establish how agents and merchants can work together to make that happen.
5
226
this is very useful for everyone building around agentic commerce
Running a few experiments on muse ember, will definitely this
2
148
this is probably the best thing happened this Diwali, I have not opened my laptop in last week and still 2x my output
In the start of the year, I moved the whole setup on a server.
I’m running all sorts of automation here, from client video updates to my software factory.
And now a year of compounding automations are accessible from my grok bot.
Grok Bot can now join your personal or company tailnets on Tailscale.
Use it to securely reach private MCP servers, internal tools, and apps on your network.
5
12
502
all the roads lead to a very agentic commerce world
Introducing Reddit Reviews: your shopping tabs’ group chat.
Reddit product conversations, organized into reviews with key community takeaways and links to the original discussions.
Rolling out gradually to US users on web, iOS, and Android. bit.ly/4C0cQom
4
170
this is huge, things are changing super fast for storefronts. Happy to build in Agentic commerce space.
machinecommerce.co
🚨 𝗝𝗨𝗦𝗧 𝗜𝗡: @Mastercard has signed a broad strategic partnership with @OpenAI.
Agentic commerce just got another major infrastructure push.
The partnership includes Mastercard’s Agent Pay, but CEO @MiebachMichael says it goes much further, covering payments, services growth, and AI transformation.
Mastercard is building around four requirements for agentic commerce:
→ Trust in the payment
→ Trust in the AI agent
→ Trust in the ecosystem
→ Merchant reach
And it now has products targeting each layer:
👉 Agent Pay — tokenized payments for AI agents
👉 Verifiable Intent — captures what the buyer actually authorized
👉 Agentic Trust Intelligence — helps banks and merchants assess AI-initiated transactions
👉 Agent Connect — helps merchants become discoverable to AI agents
Shopify is already among the first major deployments, with Shop Pay using Agent Pay and network tokens.
And Mastercard is making one important bet:
𝗧𝗼𝗸𝗲𝗻𝘀, 𝗻𝗼𝘁 𝗼𝗻𝗲-𝘁𝗶𝗺𝗲 𝘃𝗶𝗿𝘁𝘂𝗮𝗹 𝗰𝗮𝗿𝗱 𝗻𝘂𝗺𝗯𝗲𝗿𝘀.
Miebach says agentic commerce is still in its “early days,” but added that agentic transactions are already “ramping up at impressive weekly growth rates.”
Holiday shopping could be the first big test.
The battle to become the payment infrastructure for AI agents is getting VERY real.
Visa, Mastercard, Stripe, PayPal and others are all moving fast.
Who do you think is best positioned to win the agentic payments race?
1
7
338
We found a lot of interesting patterns across different shopping intents.
This chart shows how often the agent succeeded for each intent and category, across 100k runs
We let an AI agent go shopping 101,416 times.
1,044 stores. 25 categories. Every step recorded.
It finished the job about 7 times out of 10. Most failures came at checkout. Sometimes it bought the wrong thing because it never asked a question. Sometimes it gave up when it could have finished.
We wanted to see how agents really shop, and where they break.
Now we're releasing all of it for free. It's called CartTrace-100k.
1
13
465
pratham bhatnagar retweeted
What happens when an AI agent goes shopping 100,000 times?
Today we're releasing CartTrace-100k: 101,416 agent traces across 1,044 stand-in stores, every step recorded.
It got there 7 times in 10. Checkout and payment almost never broke. It got lost on the product page.
The full dataset is on Hugging Face:
huggingface.co/datasets/gene…
Built by General Machines:
generalmachines.ai
6
216
















