Technical

AnyStack ACP Agentic Commerce

Jun 19, 2026|10 min read

AnyStack now speaks to AI agents and human interfaces from the same foundation. ACP-compatible services expose your commerce capabilities to agentic clients, complementing—not replacing—the GraphQL and REST interfaces your teams already build on.

Share
Author
Raul Quintanilla
Vice President, Technology

Overview

For most of the web's history, the consumer of a commerce API was a human-facing interface: a storefront, a mobile app, an admin tool. That assumption is breaking. More and more, the consumer is an AI agent — completing a purchase, checking availability, managing a reorder, or guiding a customer through a decision on their behalf.

AnyStack is built for both. Our ACP-compatible services expose your commerce capabilities to agentic clients, sitting alongside the GraphQL and REST interfaces your teams already use, not in place of them.

What ACP is

What ACP is

The Agentic Commerce Protocol (ACP) is an open standard, originated by Stripe and OpenAI, that defines how AI agents discover products, initiate checkout, and complete transactions on behalf of buyers. It's REST- and MCP-compatible, designed to work with existing commerce infrastructure rather than forcing a rebuild. A business controls exactly which capabilities are exposed to agents, and on what terms.

In short: ACP is the plumbing that lets an AI agent become a buyer, safely, without the merchant losing control of the experience.

What ACP is

One foundation, three ways to consume it

One foundation, three ways to consume it

The core idea is simple: write the commerce logic once, expose it everywhere.

AnyStack's Backend-for-Frontend layer surfaces the same underlying capabilities through several protocols, all drawn from one set of platform adapters:

  • GraphQL — for your web and mobile UIs, via Apollo Federation. The typed, unified data graph your front-end teams already build against.

  • REST — for integrations and services that prefer a conventional interface.

  • MCP (Model Context Protocol) — for AI agents, ACP-compatible clients, and agentic workflows that need to act, not just read.

Because all three sit on the same tool implementations, a React component and an AI agent invoke the same product search, the same cart logic, the same content retrieval. One source of truth, no duplicated business logic, no separate "agent backend" to keep in sync.

One foundation, three ways to consume it

What agents can actually do

What agents can actually do

AnyStack organizes its agent-facing capabilities into clear domains, each exposing the operations an agent needs to compose a real workflow:

  • Catalog — product search, detail retrieval, recommendations, availability, context-aware pricing.

  • Cart & checkout — add and remove items, apply promotions, retrieve shipping options, initiate checkout.

  • Content — structured page and component retrieval, content search, navigation, all provider-agnostic.

  • Search & discovery — semantic search, faceting, personalization, trending.

  • Customer & orders — profiles, order history, order detail, returns, preferences.

These aren't a parallel universe of agent-only endpoints. They're the same capabilities your storefront uses, made consumable by an agent that can reason over them and chain them into multi-step actions.

What agents can actually do

Built to the standard, by design

Built to the standard, by design

ACP compatibility isn't a marketing checkbox for us. It maps to concrete architecture choices already in AnyStack:

  • REST and MCP exposure from shared tool implementations.

  • Structured, validated schemas so agents get predictable, machine-readable contracts.

  • Identity and authentication per agent, so you know who is acting and with what permissions.

  • Idempotency on mutating operations, so a retried action doesn't double-charge or double-order.

  • Event-driven webhooks for cart, order, and inventory changes.

  • Rate limiting per agent, to protect your systems at the edge.

Built to the standard, by design

Provider-agnostic, like everything in AnyStack

Provider-agnostic, like everything in AnyStack

The agentic layer inherits AnyStack's adapter model. Your commerce engine might be commercetools, Shopify, or Elastic Path; your CMS Contentful, Amplience, or Builder.io; your search Algolia or Constructor. Swapping a provider means a new adapter, not a new agent integration. The tool surface your agents call stays exactly the same.

Provider-agnostic, like everything in AnyStack

Why this matters now

Why this matters now

Agentic commerce is early. But the infrastructure decisions you make today determine whether you're ready when customers start shopping through agents instead of browsers. AnyStack gets you there without a rip-and-replace: keep your GraphQL and REST surfaces exactly as they are, and add agent-readiness as a layer on the same foundation.

That's composable architecture earning its keep: one build, ready for the interfaces of today and the agents of tomorrow.

If you're thinking about where agentic commerce fits in your roadmap, we'd welcome the conversation.

Composable Solutions. Experts Only.

Why this matters now
Raul Quintanilla portrait

Written by

Raul Quintanilla

Vice President, Technology

Raul leads technology strategy across New Elevation engagements. He focuses on composable architecture, headless content modeling, and the engineering decisions that compound — from CMS replatforms to the AEO substrate that decides whether an answer engine cites you.

More Insights

Related Reading

View All Posts →
Meet the TechWright: Senior Engineering, Amplified by AI
Blog
Technical

Meet the TechWright: Senior Engineering, Amplified by AI

AI hasn't made expertise less valuable—it's made it more decisive. The New Elevation TechWright is a senior engineer with AI woven into the delivery loop, giving you augmented development talent that moves faster without trading away judgment.

Introducing AnyStack
Blog
News

Introducing AnyStack

AnyStack is the umbrella for New Elevation's accelerator family—AnyStore, AnyPlatform, AnyNet, and AnyKit—a curated, composable foundation for building enterprise digital experiences at speed without locking you into a monolith.

Optimizing Adobe Experience Manager for IT client
Case Study
Consulting

Optimizing Adobe Experience Manager for IT client

A fortune 500 company needed clarity on whether their existing AEM implementation was optimized and whether a cloud migration made sense. New Elevation conducted a comprehensive audit — spanning templates, workflows, and stakeholder interviews — and delivered actionable recommendations that gave the client the confidence to make an informed migration decision.

Improving website performance for a leading clothing retailer
Case Study
Retail

Improving website performance for a leading clothing retailer

A leading clothing retailer's site performance issues were hurting user experience and limiting their ability to scale. New Elevation delivered a Progressive Web Application across the homepage, search, and PLP using a headless architecture, improving performance, scalability, and the foundation for future technology integration.

A Complete League Management System Without Enterprise Overhead
Case Study
Retail

A Complete League Management System Without Enterprise Overhead

A small business that runs Mahjong leagues and an online storefront needed a custom application to manage its operations without adding significant cost. New Elevation helped turn its vision into a production-ready MVP in about three weeks, enabling it to streamline league management and create a foundation for a new product offering.

Modernizing Healthcare Technology: How New Elevation Streamlined Operations and Met Critical Deadlines
Case Study
Healthcare

Modernizing Healthcare Technology: How New Elevation Streamlined Operations and Met Critical Deadlines

A healthcare provider operating multiple clinics in the Maryland area approached New Elevation with a clear need: to modernize the website and automate key integrations to improve operational efficiency and patient experience. The client required a cost-effective, timely solution that addressed both technical and healthcare-specific requirements.

We use cookies to analyze site traffic and improve your experience. By accepting, you consent to our use of analytics cookies. Privacy Policy