Agentic commerce is...actually...starting... You can now buy [coffee beans] via an AI agent using @stripe.
1/ Find [coffee] with AI
2/ Buy in 1-click with Link
3/ Agent gets one-time virtual card
4/ Agent purchases via @browserbase
☕ Watch this video (built by @EvanFenster)
This "Taste" Skill is cracked.
I can't believe I didn't discover it sooner - everyone should install this.
It works directly inside Claude Code, Codex, Hermes & more to completely kill AI-generated slop.
If you send this prompt to your agent, it will automatically install it:
@jamwt Making sure crawlers can read your website so your products get referenced in chatbots might outweigh any of the benefits SPAs provide (for any public facing pages)
@siddybuckets0 The infra setup took some time, but because a lot of this logic is built on top of our existing 100M+ product graph, not as long as it should've!
Your customers are going to ChatGPT and Gemini for their next shopping decision. Most brands have no idea what AI is saying about them.
When building our universal product catalog, we spent a ton of time figuring out how best to structure messy product data into a schema AI can understand and recommend. We built it for developers. When brands heard about this, they started asking for a way to plug directly into our system, so they could see the work we're doing on their catalogs and distribute it to ChatGPT, Gemini, and the long tail of AI shopping platforms.
@staysaasy Signing a non-compete contract w/ OAI wouldn't prevent Anthropic from steamrolling you
If Harvey/Legora had done that and only used OAI models, it wouldn't change the fact Claude for legal is a major threat
@AshConnell@threejs and im assuming some finite state machine for the characters and path finding? or how do they pick when and which functions to call?
WOW. I’m just now finding out Stripe presented my project, Barista Agent, at their keynote conference Stripe Sessions last April.
Would’ve been nice if they’d accredited me during it tho 🤨
Cursor subagents often sneaks in prompts to output entire contents of files. 1. Why is this ever necessary, and 2. why spend tokens asking a model to do this, when it can be done in half a second and no token spend with simple I/O?
Feels like a way to burn large amounts of tokens and get users to upgrade their plans quicker ...
It's never been easier to build an agentic commerce app. Use our skill to instantly get setup with product discovery and built in monetization.
`npx skills add channel3-ai/skills`
@nlevine19 I agree with a lot of this. I think of x402 as a shared, global pay-as-you-go API key that an agent can use (rather than make a new API key with each service). And this can make sense for new services, but cards work a lot better with existing businesses.
I've built agents that autonomously make purchases. The bottleneck is NOT in programmatic spending - existing card infra actually handles this very well.
There's nearly no difference between giving your employees a corporate card with complex spend limits, and doing the same thing with an agent.
Tokenizing and issuing one-time use cards is a solved problem. Amazon has been using payment tokens for years (notice how your card is saved on file, without needing to retype cvc for orders?).
The current bottleneck is being able to actually USE these cards with merchants. The current options are (1) browser use PAN form filling and (2) direct integrations (ACP, UCP, x402, etc).
You may be able to get a few large merchants with direct integrations (ACP w/ Walmart, Etsy, Instacart) but the long tail will lag, probably forever.
x402 makes sense as a global, pay-as-you-go API key, but I agree it won't fill this hole with existing merchants.
Browser use models will progress enough for the one-time use card filling approach, but there's a whole host of existing bot prevention infra that would need to be reworked for accept agents making payments as well.
Lots of interesting work to be done within this gap!
Penny combines the Channel3 API with MCP App SDK for a new, conversational shopping experience, complete with price tracking and affiliate monetization
Penny is a new MCP App that leverages the Channel3 API for product discovery, price tracking, variant data and more.
It's live in @sciraai today and rolling out to other platforms soon.
Winter Storm Fern has descended upon NJ this morning and my @yarboglobal autonomous snow blower went into action.
This is going to be a great test to see if this robot can handle a 6,000 sq.ft. driveway during a major winter storm.
I'm inside sipping a coffee while it's doing its job and so far so good!