Releases·Tool Release
PayPal engineers demo agent pre-authorization approval token days from shipping
A YouTube talk by PayPal's Jay Mok and Ben Coumes demonstrates an inverted authorization flow for AI agents: the user approves a payment before the agent has found an item or picked a merchant. PayPal returns a JSON payload specifying amount, expiry, and permitted merchant. The feature was days from shipping when recorded, signaling how agent commerce may handle human sign-off.
7 engagement1 sourceTue, Sep 1, 2026, 06:30 PM
AI Overview
- The approval token inverts the traditional synchronous PayPal order flow of find-item, open app, approve
- The returned JSON payload carries the amount, the expiry, and the merchant the agent is permitted to transact with
- The talk frames agent authorization around three questions: did the human authorize this, is it allowed right now, and accountability
- The feature was recorded days before shipping
Related
Social·Community
Stripe's Anna Spysz demos how agent persona prompts shape agentic shopping behavior
In a YouTube talk, Stripe's Anna Spysz recounts how her shopping agent pushed an aggressive upsell on headphones and got snarky when she hesitated. The cause was her own config: a persona system prompt defining the agent as an aggressive audio gear salesman. Swapping it for a patient mentor persona changed the conversation entirely, using the same tools and protocol.
In a YouTube talk, Stripe's Anna Spysz recounts how her shopping agent pushed an aggressive upsell on headphones and got snarky when she hesitated. The cause was her own config: a persona system prompt defining the agent as an aggressive audio gear salesman. Swapping it for a patient mentor persona changed the conversation entirely, using the same tools and protocol.
23 engagement1 sourceyoutubeTue, Sep 1, 2026, 04:30 PM
News·Company News
OpenAI publishes article on x402 agentic commerce payment layer
OpenAI published an article about x402, a payment protocol for agentic commerce, signaling mainstream adoption. The post highlights that OpenAI, AWS, Coinbase, and Stripe are converging on x402 as the agent payment layer. The security model restricts agents to pay only pre-approved merchants via a whitelist.
OpenAI published an article about x402, a payment protocol for agentic commerce, signaling mainstream adoption. The post highlights that OpenAI, AWS, Coinbase, and Stripe are converging on x402 as the agent payment layer. The security model restricts agents to pay only pre-approved merchants via a whitelist.
16 engagement1 sourceredditMon, Aug 17, 2026, 09:06 AM
Products·launched
Authoryze: Payment controls for AI agents
Authoryze provides payment controls for AI agents, allowing users to set spending limits and approve transactions. It solves the problem of unauthorized or excessive spending by autonomous agents.
Authoryze provides payment controls for AI agents, allowing users to set spending limits and approve transactions. It solves the problem of unauthorized or excessive spending by autonomous agents.
5 engagement1 sourcehackernewsSun, Aug 2, 2026, 08:34 PM
Social·Community
Community explores what autonomous payments mean in the AI agent economy
A Reddit user asks what autonomous payments really mean in the AI agent economy, arguing it is not binary. They describe a spectrum from per-purchase approval to broad delegated authority, with agents discovering services, comparing options, executing payments, and receiving results. The post notes that most current activity is agent-to-service/API payments, citing x402 and MPP.
A Reddit user asks what autonomous payments really mean in the AI agent economy, arguing it is not binary. They describe a spectrum from per-purchase approval to broad delegated authority, with agents discovering services, comparing options, executing payments, and receiving results. The post notes that most current activity is agent-to-service/API payments, citing x402 and MPP.
9 engagement2 sourcesredditSun, Aug 23, 2026, 01:00 PM
Social·Community
Discussion: who holds authority when an AI agent operates across multiple systems?
A Reddit discussion asks who is responsible for authority when an AI agent's actions chain through multiple systems — identity providers, CRMs, internal APIs, third-party models, payment systems, and approval workflows. Each component has its own permissions, but no clear owner governs the full chain. The thread highlights an unresolved governance gap for practitioners deploying multi-system agents.
A Reddit discussion asks who is responsible for authority when an AI agent's actions chain through multiple systems — identity providers, CRMs, internal APIs, third-party models, payment systems, and approval workflows. Each component has its own permissions, but no clear owner governs the full chain. The thread highlights an unresolved governance gap for practitioners deploying multi-system agents.
37 engagement1 sourceredditMon, Aug 31, 2026, 08:39 PM