Rye
Also known as: Rye.com, Rye Universal Checkout, Sell Anything API
Agentic commerce infrastructure whose Universal Checkout API turns a product URL and tokenized payment into a placed order on the merchant's site, with a confirmation hold, price limits, and returns and product data APIs.
Rye is agentic commerce infrastructure: its Universal Checkout API turns a product URL and a tokenized payment method into a placed order on the merchant's own site. A developer creates a checkout intent; RyeBot, Rye's web agent, retrieves the live offer with real price, tax and shipping, and the intent waits for confirmation so the buyer can see the final cost before anything is charged.
On confirmation Rye completes the checkout through AI browser automation and returns the order status, with a single step Purchase endpoint for fire and forget flows. Maximum price thresholds keep orders within budget, card data stays tokenized and never reaches the agent, and merchants remain merchant of record. Direct routes cover Amazon, Shopify and Best Buy, and the automation works across other web stores; Rye caches successful flows for speed. A Returns API files returns and tracks refunds, and a Product Data API reads product details from most product URLs.
The API has SDKs, webhooks, separate staging and production environments and a docs MCP server. Current limits include US only ordering (other regions on request for enterprise), physical goods, guest checkout and one product per checkout. The console has no team accounts, and the trust center could not be read.
Rye was founded in 2021 and moved into agentic commerce infrastructure in 2025.
Vendor details
Canonical URL
https://rye.com
Category
Agent infrastructure
Subcategory
Universal checkout execution layer for agentic commerce
Funding status
Not stated on the first party pages read.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
The Universal Checkout API places orders on any web store reachable by a product URL, with direct routes for Amazon, Shopify and Best Buy; payments use tokenized cards from Stripe and other gateways, plus x402 payments through AgentCash; the Returns API files returns with merchants and the Product Data API reads product details from URLs. SDKs for npm and pip, webhooks and a docs MCP server support integration.
In practice
An AI shopping app creates a checkout intent from a product URL, shows the buyer Rye's resolved price, tax and shipping, and confirms with a tokenized card to place the order
An agent is allowed to buy only below a maximum price, so Rye places nothing that exceeds the budget
A shopper returns an agent bought item through the Returns API and the app follows the return to the refund
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
5.0 / 14 capabilities · 36%
| Integrations & Tool Calling | Full |
|---|---|
|
Agents take authenticated action in outside systems. The Universal Checkout API places a real order on any web store reachable by a product URL, with direct routes for Amazon, Shopify and Best Buy (Best Buy after approval), pays through tokenized cards from Stripe and other gateways, and files returns with the merchant through the Returns API; the Product Data API reads product details from most product URLs. SourceRye, rye.com/docs llms-full.txt (API Comparison, Submit Orders to Best Buy, Returns, Confirm checkout intent) and rye.comread 2026-09-25 |
|
| Workflow Orchestration | Partial |
|
A fixed lifecycle the customer steps through, not a workflow model. A Checkout Intent moves through retrieving_offer, awaiting_confirmation, placing_order and completed or failed, and a single step Purchase endpoint runs offer retrieval, payment authorization and order placement asynchronously. The customer cannot compose steps, branches or agents; the sequence is Rye's checkout. SourceRye, rye.com/docs llms-full.txt (Checkout Intent Lifecycle, Single Step Checkout)read 2026-09-25 |
|
| Knowledge Grounding & RAG | Not documented |
|
No retrieval structure over the customer's knowledge. The Product Data API returns details for a product URL on request, which is a lookup of merchants' public pages, not a maintained index of the customer's corpus. SourceRye, rye.com/docs llms-full.txt (Product Data)read 2026-09-25 |
|
| Human Oversight & Guardrails | Full |
|
Rye holds the order until someone confirms it. A Checkout Intent stops at awaiting_confirmation once the live offer (price, tax, shipping) is ready, and Rye places nothing until the intent is confirmed with payment, so a buyer can see final landed cost before the charge; the single step endpoint is the opt out for fire and forget flows. Maximum price thresholds mean an order is only placed within budget. SourceRye, rye.com/docs llms-full.txt (Checkout Intent Lifecycle, Confirm checkout intent, Price Constraints)read 2026-09-25 |
|
| Security, Identity & Governance | Not documented |
|
There is no team access model: the docs say the Rye console does not support team accounts and suggest sharing one login through a password manager, so there is no role surface (Google sign in only authenticates a single account). The homepage's security points (tokenized payments only, merchants stay merchant of record, PCI DSS scope reduction for developers) are payment handling, not an attestation of Rye's own. A trust center sits at trust.rye.com, but its contents load only in a browser, so which attestations Rye holds is unconfirmed. SourceRye, rye.com/docs llms-full.txt (Account Registration, Team accounts), rye.com (Security and Compliance) and trust.rye.comread 2026-09-25 |
|
| Observability & Auditability | Partial |
|
State per order, not a trace of the agent's work. Each checkout intent and return exposes its current state through the API, and webhooks send thin event notifications with deterministic ids so the latest state can be fetched. That shows where an order stands, not what RyeBot did step by step on the merchant site. SourceRye, rye.com/docs llms-full.txt (Webhooks, Checkout Intent Lifecycle, Returns)read 2026-09-25 |
|
| Memory & State Persistence | Not documented |
|
Rye caches successful checkout flows as deterministic workflows for speed, but that is a replay cache the automation applies mechanically, not memory an agent keeps. Promo code caching is cost optimization. SourceRye, rye.com (home) and rye.com/docs llms-full.txt (Promo Codes)read 2026-09-25 |
|
| Deployment & Data Residency | Not documented |
|
Hosted API with no residency or customer environment option. Staging and production are separate, isolated environments of Rye's hosted service, not deployment choices; ordering is US only, with other regions on request for enterprise, which is where goods ship, not where data lives. SourceRye, rye.com/docs llms-full.txt (Environments, API Limitations)read 2026-09-25 |
|
| Prebuilt Agents, Templates & Packs | Not documented |
|
API primitives, not prebuilt agents. The docs give example flows and tutorials (a chat storefront with Stripe in Next.js, an app built with Lovable) and a config that turns a coding assistant into a Rye integration helper; those are guides the customer builds from. The site also lists Rye in ChatGPT and Rye for OpenClaw; neither is described as an agent a buyer adopts. SourceRye, rye.com/docs llms-full.txt (example flows) and rye.com (products menu)read 2026-09-25 |
|
| Triggers & Channel Coverage | Not documented |
|
Work begins only when the caller asks: a checkout starts when the developer's app creates or submits a checkout intent, and webhooks go outward to report state without starting work. SourceRye, rye.com/docs llms-full.txt (Checkout Intent Lifecycle, Webhooks)read 2026-09-25 |
|
| Model Flexibility & Routing | Not documented |
|
No model choice. Rye's AI browser automation runs on models Rye selects and does not name; the customer calls a checkout API and never picks or routes a model. SourceRye, rye.com (home) and rye.com/docs llms-full.txtread 2026-09-25 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
A documented public API for Rye's own platform. The REST API (checkout intents, purchase, products, returns, payment gateway tokenization credentials) is published with an OpenAPI reference, API key auth, rate limits, error handling and staging and production environments, with SDKs (checkout-intents for npm and pip), webhooks, and a docs MCP server at docs.rye.com/mcp. SourceRye, rye.com/docs llms.txt and llms-full.txt (API reference, Authorization, Environments, Webhooks)read 2026-09-25 |
|
| Testing, Debugging & Optimization | Not documented |
|
No evaluation surface. The staging environment lets a developer test an integration without real orders, which tests the customer's own code against Rye, not an agent's behavior; the homepage reliability and latency figures are the vendor's own performance claims. SourceRye, rye.com/docs llms-full.txt (Environments) and rye.comread 2026-09-25 |
|
| Browser & Computer Use | Full |
|
Rye's own agent drives real merchant checkouts. RyeBot is Rye's web agent that retrieves offers and performs checkout flows on merchant sites on behalf of shoppers (merchants can allowlist it by its user agent), and the homepage describes AI browser automation with fraud mitigation working across any web store. SourceRye, rye.com/docs llms-full.txt (RyeBot Allowlisting) and rye.com (home)read 2026-09-25 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Developer plan at $149 per month including fifteen hundred orders, then $0.05 per additional order, with a thirty day free trial
monthly subscription with included order volume plus per order overage; separately per request pricing for x402 stablecoin payments
What is public
Developer plan price, monthly credit, per order and per product fetch fees, the free trial and the Enterprise inclusions.
Billing mechanics
Monthly Developer subscription with $50 in credits, then $0.05 per order placed and $0.02 per product fetch; Enterprise volume pricing.
Cost watchouts
Per order and per product fetch fees sit on top of the monthly plan once the $50 credit is used, and third party platform costs pass through; x402 payments through AgentCash are priced per request separately.
Variable cost rationale
Cost scales directly with order volume: the Developer plan includes fifteen hundred orders and charges $0.05 per additional order, and on the x402 path each intent and confirmation carries a per request fee, with product price and payment processing passing through on top. Spend tracks transactions placed, so a busy agent's bill grows with usage.
Additional watchouts
The $149 plan covers fifteen hundred orders; product price and payment processing are separate pass through costs, so total spend depends heavily on order volume and average order value, not the subscription alone.
Sales call required
Mixed (some tiers require a call)
Free / trial
Thirty day free trial on the Developer plan
Lowest paid plan
Developer, $149/month
Key ambiguities
The plan changed since July: the 1,500 included orders are gone, replaced by $50 in monthly credits and per call fees.
Missing data
Enterprise tier rates and volume thresholds are not published, and per order economics at very high volume beyond the $0.05 overage are not detailed.
Related vendors
- AgentOps — Agent observability and debugging platform: open source SDKs trace…
- Agno — Python agent framework and AgentOS runtime (formerly Phidata) for…
- AIsa — Resource and payment gateway for AI agents: one key to 110+ models…
- AlphaBitCore — AI control plane for regulated financial firms: one gateway enforces…
- Anchor Browser — Cloud hosted browser infrastructure that lets AI agents operate real…
- Apify — Cloud platform and marketplace of more than 73,000 ready-to-run…
Alternatives to Rye
The closest documented capability profiles to Rye among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Skyfire4.5 / 14Adds documented Security, Identity & Governance
- Lightpanda6.0 / 14Adds documented Deployment & Data Residency and Model Flexibility & Routing
- Browserless8.5 / 14Adds documented Security, Identity & Governance and Deployment & Data Residency, among others
- Catalog4.5 / 14Adds documented Knowledge Grounding & RAG and Triggers & Channel Coverage
- Circuit & Chisel4.5 / 14Adds documented Memory & State Persistence and Model Flexibility & RoutingRye vs Circuit & Chisel →
- Kernel8.5 / 14Adds documented Security, Identity & Governance and Deployment & Data Residency, among others
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded