Agentic Commerce Infrastructure Startups — Skyfire, Basis Theory, Nekuda, and Rye

Who Is Building the Infrastructure for Agentic Commerce

According to Visa, AI agent traffic to retail sites has surged 1,200% year-over-year. ChatGPT alone processes approximately 50 million shopping-related queries per day. Yet the conversion rate of this traffic into actual purchases remains extremely low. The reason is straightforward: the infrastructure for AI agents to "go shopping" does not yet exist.

Viewing the agentic commerce ecosystem through the 7-layer value chain, major tech companies are populating the AI platform and protocol layers. Google's UCP, OpenAI's ACP, and Visa's TAP are making headlines. But who is building the "plumbing" underneath — the infrastructure that lets agents handle card data, add products to carts on merchant sites, and complete payments?

That question is what the four startups in this article address. Rather than presenting them as an equal catalog, this analysis starts from the challenges at each infrastructure layer and positions each company as a solution to a specific problem.

The Payment Data Wall — The Problem Basis Theory Is Solving with $33 Million

The biggest technical challenge in agentic commerce is deceptively simple: how do you safely pass a credit card number to an AI agent?

When humans shop on e-commerce sites, they enter card numbers into browser payment forms. This data is processed under PCI DSS (Payment Card Industry Data Security Standard), a rigorous security framework. But when an AI agent shops on behalf of a user, where is the card data stored, who manages access, and how is it transmitted to the merchant? Passing card numbers directly to a large language model (LLM) is out of the question.

In October 2025, Basis Theory raised $33 million in Series B funding precisely because of this problem. The company was originally founded in 2020 as a PCI-compliant tokenization platform, converting merchant card data into tokens (encrypted substitute values) for safe storage and use.

This technology is being redefined for the agentic commerce era. In June 2025, Basis Theory launched basistheory.ai and established the Agentic Commerce Consortium. This consortium, with over 20 member companies, is developing an open framework that lets merchants opt into accepting agent-initiated transactions.

Costanoa Ventures' Amy Cheetham noted that "Basis Theory enables AI agents to authorize transactions, personalize experiences, and drive autonomous purchasing." Notably, the company is complementary rather than competitive with existing token services from Visa and Mastercard. It normalizes card network tokens into a format that agents can access in real time with scoped permissions — essentially, a token on top of a token.

The structural payment gap identified by a16z crypto exists on the established merchant side as well. Basis Theory's promise to reduce PCI compliance burden by up to 90% is valued as a payment data management efficiency play even before agentic commerce. The agentic wave adds a new growth driver to this existing value proposition.

Building an "ID Card" for AI Agents — Skyfire

Even after solving safe payment data transfer, a fundamental question remains. Is the AI agent accessing a merchant's website truly acting on behalf of a legitimate user? Or is it a malicious bot?

Skyfire confronts this "trust problem" head-on. Founded in 2023 by Amir Sarhangi and Craig DeWitt, the company has raised $9.5 million in total. The fact that a16z CSX and Coinbase Ventures joined a later round signals that crypto and payments infrastructure investors recognize the criticality of this problem.

Skyfire's core technology is the KYA (Know Your Agent) protocol. Named after the financial industry's KYC (Know Your Customer), this framework issues "identification documents" for AI agents. It uses standard JSON Web Tokens (JWTs), maintaining compatibility with existing OAuth2 and HTTP infrastructure, so merchants require no major system overhauls.

Why does Skyfire's approach deserve particular attention among these four companies? The March 2026 partnership with F5 provides the answer. F5 is a global leader in application delivery and security, and its Distributed Cloud Bot Defense protects many e-commerce sites. Integrating Skyfire's KYA tokens into F5's bot defense infrastructure enables the precise traffic control of "let authenticated AI agents through, stop malicious bots." This integration is scheduled for availability by late April 2026.

Furthermore, in December 2025, Skyfire partnered with Visa's Intelligent Commerce Suite and Trusted Agent Protocol to demonstrate a flow where an agent evaluates products on Consumer Reports and completes a purchase on Bose.com. The significance lies in demonstrating a consistent flow from identity authentication to payment completion on existing card networks.

The Blank Spot in Commerce Execution — Nekuda and Rye

The payment data and identity layers examined so far establish the foundation of "trust" for transactions. But even with trust established, transactions cannot complete without the infrastructure for AI agents to actually select products, add them to carts, and finish checkout. As Rye's analysis points out, "surprisingly few pure-play companies" operate in this checkout execution layer.

Nekuda and Rye are attacking this blank spot with different approaches.

Nekuda solves the problem from the merchant side. Founded in 2024, the company raised $5 million led by Madrona, with Amex Ventures and Visa Ventures participating. The simultaneous investment from two major payment networks is noteworthy. Nekuda provides structured commerce API endpoints for merchants to expose to AI agents — catalog search (/catalog), cart management (/cart), checkout (/checkout), and order tracking (/post-order) — enabling agents to complete transactions without URL redirects.

The founders described this design philosophy as "headless commerce for a headless internet" on Substack. While traditional headless commerce separated the human frontend from the backend, Nekuda's APIs eliminate the need for a human frontend entirely. The architecture is built for a world where AI agents call APIs directly to complete purchases.

Rye, in contrast, does not wait for merchant-side adoption. Founded in 2021 by Justin Kan (Twitch co-founder) and Arjun Bhargava (ex-Reddit), the company raised $14 million led by a16z crypto. Its Universal Checkout API takes a product URL and payment token, then completes purchases on any e-commerce site without prior merchant integration.

Technically, it combines an AI browser agent, deterministic workflow caching, DOM normalization, and a fraud-mitigation proxy layer. Published SLAs show Amazon reliability at 99% with 5-second latency, and Shopify at 96% with 20-second latency. For AI flows on non-integrated sites, reliability stands at 65% with 5-minute latency — a number honestly disclosed as still maturing.

The difference between these two companies also reflects different views on the agentic commerce "transition period." Nekuda provides infrastructure for a future where merchants proactively expose agent-friendly APIs, while Rye enables agent purchases today, even when merchants have not yet adapted. They are more complementary than competitive, each suited to different stages of market maturity.

What VC Funding Patterns Reveal About Market Dynamics

Surveying startup funding across agentic commerce infrastructure, including these four companies, a clear pattern emerges.

Basis Theory's $33 million and Skyfire's $9.5 million total $42.5 million for the payments and identity layer alone. Adding Nekuda's $5 million brings the total to approximately $47.5 million. Meanwhile, Rye, focused on checkout execution, raised $14 million. On the merchant enablement layer, FERMAT raised a $45 million Series B.

What does this distribution mean? VCs are placing the largest bets on two ends of the spectrum: "enabling agents to spend money safely" and "enabling merchants to sell to agents." Viewed alongside moves by established players and protocols like Stripe's Shared Payment Token (SPT) and x402's machine payment rails, it becomes clear that the payments layer is recognized market-wide as the biggest bottleneck.

The investor roster tells another interesting story. Nekuda's backers include both Amex Ventures and Visa Ventures. Skyfire attracted a16z CSX and Coinbase Ventures. Basis Theory counts Bessemer Venture Partners and Kindred Ventures among its investors. Payment networks, crypto firms, and top-tier VCs are each betting on this space from their own vantage points.

The simultaneous investment by Visa Ventures and Amex Ventures in Nekuda is particularly telling. When VC arms of two competing card networks fund the same startup, it signals a shared recognition that "unless someone standardizes the commerce execution layer for agentic commerce, transaction volume growth for the card networks themselves will be constrained."

Summary

Agentic commerce infrastructure is rapidly taking shape: Basis Theory tokenizes payment data, Skyfire handles identity and authentication, Nekuda provides merchant-facing commerce APIs, and Rye delivers universal checkout. Combined, approximately $62 million has been invested in this "plumbing."

The real transformation, however, comes not from the funding total but from the moment these infrastructure pieces actually connect. Skyfire's KYA token passes through F5's bot defense, Basis Theory's token safely transfers payment credentials, Nekuda's API assembles the cart, and Rye's engine completes checkout. The moment this pipeline operates end-to-end will mark the inflection point for agentic commerce in practice.