Agentic AI Is Moving From Hype to Checkout Button
Agentic AI in commerce refers to artificial intelligence systems that can independently discover products, answer questions, and complete transactions for customers across digital touchpoints, acting as always-on sales and service representatives that sit inside chat, voice, and other interfaces while connecting directly to payment and ordering platforms.
The story behind Claude Sonnet 5 is not another model launch; it is a shift in the economics of automation. Anthropic describes Sonnet 5 as a more powerful and agentic version of its midsize model that can plan, use tools like browsers and terminals, and run autonomously at a level that recently required larger, more expensive systems. At the same time, Square has turned this new generation of AI into live commerce integrations with an app for ChatGPT and a plugin for Claude to enable agentic commerce. The message is clear: AI agents are moving from experimental pilots to production channels where customers discover products and place orders inside the AI interfaces they already use.
Claude Sonnet 5 and the Cost Logic of AI Agents
Anthropic’s launch of Claude Sonnet 5 is explicitly framed around agentic capabilities becoming table stakes among foundation model companies. This is not about winning benchmarks; it is about making multi-step, tool-using agents affordable enough to run all day. Sonnet 5 can make plans and operate tools such as browsers and terminals autonomously, which matters because commerce workflows—discovery, recommendation, stock checks, order edits—are inherently multi-step and stateful. When those workflows no longer require a top-tier, premium model for every action, the business case changes. Developers can architect agents that call Sonnet 5 for most work and reserve more expensive models for rare edge cases. Anthropic is positioning Sonnet 5 at a steep discount to its top model as it races toward a blockbuster IPO, signaling that lower agentic workflow costs are now a strategic battleground, not an afterthought.
Square Turns Foundation Models Into Plug-In Commerce
Square has already moved from theory to execution by wiring ChatGPT and Claude into real ordering flows. The company has announced integrations with these platforms to help enable agentic commerce. These integrations, delivered as an app for ChatGPT and a plugin for Claude, let sellers get discovered and transact at the exact moment customers are making purchasing decisions through AI-powered conversations. Importantly, Square promises no additional marketplace commissions on orders placed through these integrations, protecting margins that are under pressure from traditional delivery platforms. For early Food & Beverage merchants with active Square Online Ordering profiles, this means that when a customer asks an AI assistant where to order lunch, the agent can display accurate menu data and complete the order directly in a supported AI experience. The seller does not pay a technical tax in the form of new APIs, setup work, or extra fees.
Square’s strategy is opinionated: sellers should not have to manage a patchwork of AI channels. Eligible sellers are opted in with no additional work, setup, or fees required, and there is no API to build or new tool to configure. Square syncs business information, menu data, hours, and ordering details in real time while it participates in emerging agentic commerce protocol groups to help shape open standards for how AI agents and commerce platforms interact. That approach treats AI interfaces like search or maps: another surface where commerce should happen automatically, not another software project every small merchant must maintain.
What This Means for Merchants and the Next Wave of Competition
For ordinary merchants, the impact is practical rather than futuristic. Square says these AI-powered conversations help sellers get discovered and transact at decision time, which translates into more orders without extra operational overhead. Brooklyn-based Partners Coffee illustrates the point: with agentic commerce and AI tools working in the background, the brand reports being digitally discovered and growing in efficiency while keeping a lo-fi, coffee-first in-store experience. This split—digital automation, offline authenticity—is what many small businesses want but could not afford to engineer. Now, with mid-range models like Claude Sonnet 5 handling tool use and platforms like Square hiding the integration complexity, AI agents commerce stops being a custom build for large enterprises and becomes a default feature of mainstream payment and commerce stacks.
The competitive landscape will adjust. As Anthropic, OpenAI, and Google converge on more agentic offerings—OpenAI’s latest GPT model supports subagents for longer tasks, while Google’s recent Gemini release is positioned as an agentic tool that plans and iterates on work with minimal human input—the differentiator will not be who has agents, but who can run them economically inside real businesses. Square’s claim that more AI integrations are coming, plus its broader strategy to give sellers visibility across new AI platforms without individual integrations, suggests that payment providers and commerce platforms that fail to embed low-friction agents will look outdated. In that context, Claude Sonnet 5 pricing strategy—and models like it—becomes a lever for smaller players to offer enterprise-grade automation without enterprise budgets.
The Takeaway: Agents Will Be Invisible, but Their Economics Will Decide Who Wins
AI agents will not feel like new apps; they will feel like the natural way to order, ask questions, and pay anywhere customers talk to machines. What will matter is which platforms can keep those conversations accurate, fast, and affordable at scale. Anthropic’s release of Claude Sonnet 5 as a more agentic midsize model at a steep discount to its flagship and Square’s move to make ChatGPT and Claude integrations opt-in, fee-free, and commission-free for eligible sellers point toward the same conclusion: the real disruption in AI agents commerce is cost, not novelty. As new AI platforms and channels emerge, Square plans to give sellers visibility without forcing them to manage each integration individually, while Anthropic and its rivals race to ship more capable yet cheaper agents. The platforms that hide complexity and keep agentic workflow costs low will set the new baseline for how commerce is automated—and who survives in the next wave of competition.
