207 companies. $2.4B deployed. Zero in Latin America.

Between Q4 2025 and Q1 2026, the agentic commerce space went from a cluster of experimental projects to a defined market category with serious capital behind it. CB Insights maps 90+ companies. agenticcommercemap.com lists 207+. More than $2.4B has been deployed across the full stack when you include infrastructure incumbents repositioning for agent-native commerce: Stripe ($14B revenue, $95B valuation), Marqeta ($550M revenue, public), Brex ($1.2B+ raised), alongside $70M+ in new-entrant funding at the payments and identity layer alone. Six competing protocols have been announced. And the category has its own conference circuit, its own analyst reports, and its own vocabulary.

We built this map because we needed it. Before deciding what to build and how to position it, we had to understand who was already building what, which layers were crowded, which were empty, and where genuine white space existed. What we found reshaped our entire strategy.

This is not just a market map. This is a strategic analysis. We will tell you who is funded, how much, by whom, and what it means. We will tell you which categories are defensible and which will be commoditized. We will tell you where we think the entire space is going, and why Latin America is the best market to build in right now.

The 207 companies, broken down

The agenticcommercemap.com database lists 207+ companies as of April 2026. We analyzed every one. Here is how they break down by category, funding stage, and geography.

By category

The distribution reveals the market's center of gravity: AI platforms and tool providers account for 45% of all companies. This is the crowded center. Payments and identity (16%) is where the money is flowing. Checkout execution (11%) is where the product differentiation lives. Everything else is supporting infrastructure.

By funding stage

Of the 207 companies, we were able to confirm funding data for 89. The rest are either bootstrapped, corporate-backed (internal teams at Google, Visa, Stripe, etc.), or pre-funding.

The funding stage distribution tells a clear story: this is an early-stage market. 70% of funded companies are at seed or Series A. Only 14 companies have reached Series B or beyond. The market belief is strong (investors are putting money in), but the market reality is still forming (most companies are pre-revenue or early revenue).

By geography

This is the number that reshaped our strategy.

69% of all agentic commerce companies are based in the United States. 15% are in Europe. 7% are in Israel. Zero are in Latin America. Zero are in Africa. The entire Global South, home to 6 billion people and representing $15+ trillion in annual GDP, has no dedicated agentic commerce infrastructure provider.

This is not because the market does not exist. Brazil alone processed R$17.4 trillion in Pix transactions in 2025 (Banco Central data). Mexico's SPEI system processes $2.7 trillion annually. LatAm digital payments are projected to exceed $2 trillion in annual volume by 2027 (Americas Market Intelligence). The market exists. The infrastructure for agents to participate in it does not.

The funding map: who raised what, from whom

Money reveals strategy. Here are the largest and most strategically significant funding rounds in agentic commerce, organized by value chain layer.

Payments & Identity (Layer 3)

This layer received over $70M in combined funding between 2024 and Q1 2026. It is the most actively funded layer in the stack because the problem it solves, how an AI agent pays for things with verified identity, is the prerequisite for everything else.

Three patterns in this data:

1. Strategic investors, not just financial. Visa appears in two rounds (Payman, Nekuda). Coinbase in one (Payman). Microsoft (via M12) in one (Basis Theory). Amex in one (Nekuda). The card networks and tech giants are placing bets across the layer, not consolidating on a single winner.

2. Crypto meets tradfi. Skyfire (a16z crypto + Circle) and Payman (Visa + Coinbase) represent the convergence of crypto infrastructure and traditional payment rails. The funding thesis: agent payments will be rail-agnostic, and companies that bridge crypto and traditional rails have an advantage.

3. All US/EU focused. Every company in this list operates exclusively in the US and/or Europe. Not one has LatAm rail integration. Not one issues NF-e. Not one connects to Pix. The $70M+ invested in this layer has purchased zero capability in the LatAm market.

Checkout Execution (Layer 5)

Composio's $50M+ round deserves special attention. Composio is not a commerce company - it is a generic tool integration platform (400+ app integrations for AI agents). But its round is the largest in the adjacent space and signals investor conviction that the “agent tool infrastructure” category is fundable at scale. Composio is building horizontal (any tool for any agent). We are building vertical (commerce tools for LatAm agents). These are fundamentally different businesses with different moats.

Trust & Security (Layer 7)

Forter and Signifyd are incumbents in e-commerce fraud prevention that are pivoting to agent transaction verification. AgenticTrust is a startup purpose-built for agent identity and trust. The strategic question: can fraud prevention incumbents adapt to agent transactions, or will purpose-built startups win? Our bet: incumbents adapt. Fraud detection is fundamentally about patterns, and the incumbents have 10+ years of transaction pattern data. But they will be slow, giving startups a 12-18 month window.

The revenue reality: funding vs. revenue across the stack

The VC reviewer's most important question about any landscape map is: who actually has revenue? Here is our best estimate for the major players, combining public filings, headcount data (LinkedIn), pricing page analysis, and market signals. Revenue estimates marked with ~ are inferred.

The pattern is clear: revenue concentrates at the infrastructure incumbents (Stripe, Marqeta, Forter, Signifyd, Brex), while the agentic-native startups are overwhelmingly pre-revenue. This is not a weakness; it is the nature of infrastructure markets. Stripe was pre-revenue for 2 years before processing its first dollar. The companies that build the right infrastructure now will capture the revenue as agent commerce volume materializes over 2027-2029.

The 7-layer framework

Every transaction an AI agent initiates travels through a stack of distinct functions. Rye's framework maps them cleanly. Understanding which layer a company operates in tells you almost everything about its business model, its moat, and its competitive dynamics.

CodeSpar operates transversally across layers 2, 3, 5, and 6 simultaneously. This is unusual. Most players anchor in one layer and go deep. The cross-layer architecture is both the thesis and the technical challenge. We chose this architecture because LatAm commerce cannot be solved at a single layer. You cannot do checkout (layer 5) without handling fiscal compliance. You cannot handle fiscal compliance without connecting to payment rails (layer 3). You cannot connect to payment rails without exposing tools via MCP (layer 2). The layers are coupled in LatAm in ways they are not in the US.

Layer-by-layer competitive analysis

Layer 3: Payments & Identity - the most crowded new entrant layer

This layer received over $70M in combined funding between 2024 and early 2026. The central problem: how does an AI agent pay for something, with verified identity and a clear mandate, in a way that is auditable and revocable? Five companies have raised significant rounds to answer this question. None of them answer it for Latin America.

Layer 5: Checkout Execution - where CodeSpar's Complete Loop lives

This is where the value chain gets interesting. Given that an agent has the intent to transact and the identity/mandate to do so, how does it actually execute the full commerce workflow? Most companies in this layer focus on the purchase moment. We focus on the complete lifecycle.

The gap matrix

Here is what each major player does and does not do. This table is the clearest way to see the white space.

The last two rows tell the story. Every company in this map operates in the US or Europe. Not one covers Latin America. Not one handles Brazilian fiscal compliance. Not because they have decided it is too small, but because they do not have the domain knowledge to build for it.

The white space: zero players in LatAm commerce for agents

This is the strategic insight that defines our positioning. Of 207+ companies building agentic commerce infrastructure, zero operate in Latin America. This is not a niche gap. This is a structural white space in a $2+ trillion market.

The LatAm digital payments market

The numbers are large enough to build a category-defining company:

Brazil alone represents one of the largest digital payment markets in the world. Pix, launched in November 2020, processed more transactions in 2025 than all US debit card transactions combined. 76% of Brazilian adults use Pix. It is not a niche payment method; it is the default financial infrastructure for 215 million people.

Mexico's SPEI (Sistema de Pagos Electronicos Interbancarios) processes $2.7 trillion annually. Colombia's PSE handles $180 billion. Argentina's Transferencias 3.0 is growing rapidly. Each of these is a real-time payment rail with unique API requirements, fiscal compliance obligations, and regulatory frameworks.

Agent commerce TAM

Estimating the agent commerce TAM requires making assumptions about adoption rates, which are inherently uncertain. Here is our framework:

Competitive moats: what is defensible and what is not

Not every category in this landscape has a moat. Some of the most-funded categories are the least defensible. Understanding moat dynamics is essential for anyone building or investing in this space.

Strong moats

Payment rails (Layer 3): Defensible because of regulatory licensing, banking partnerships, and compliance requirements. Skyfire needs crypto licenses. Nekuda needs card program manager agreements. Natural needs payment processor relationships. These take 6-18 months to establish and cannot be replicated quickly. Geographic moats are even stronger: a US payment infrastructure company cannot add Pix support by hiring one engineer. It requires Banco Central registration, a Brazilian banking partner, NF-e certification, and SEFAZ integration with 27 state authorities.

Trust & security (Layer 7): Defensible because of data network effects. Forter has processed billions of transactions and has a proprietary fraud model trained on that data. A new entrant starts with zero transaction history. This moat compounds over time: the more transactions you process, the better your fraud model, the more merchants trust you, the more transactions you process.

Regional commerce infrastructure (cross-layer): Defensible because of domain expertise accumulation. Building for LatAm commerce requires understanding 92 types of fiscal documents, 50+ carriers, country-specific ERP systems, regulatory frameworks that change quarterly, and payment rails that US developers have never heard of. This knowledge cannot be acquired from a documentation page. It is accumulated through years of building, failing, and iterating in the region.

Weak moats

Generic tool integration (Layer 2): Composio has 400+ integrations, but each one is a thin API wrapper. Any well-funded competitor can build the same wrappers in 6 months. The integration itself is not defensible; the depth of the integration is. Composio is vulnerable to vertical players (like us) that go deeper in specific domains.

Protocol abstraction: Firmly abstracts ACP, UCP, and MCP into a unified checkout. This is valuable today when protocols are fragmented. If protocols consolidate (as we predict they will by 2027), the abstraction layer becomes unnecessary. Firmly's moat is temporal, not structural.

Card issuance (Layer 4): Lithic, Marqeta, and Unit issue virtual cards to agents. This is a valuable capability but not a defensible one. Card issuance is a regulated commodity: multiple providers offer the same capability with similar pricing. The moat is in what you do with the card, not in issuing it.

CodeSpar's positioning

We are not competing with Composio. Composio integrates 400+ services superficially. We integrate 50+ LatAm commerce services deeply. We are not competing with Stripe. Stripe provides payment rails. We provide commerce orchestration on top of local payment rails. We are not competing with Rye. Rye executes US checkout. We execute the LatAm Complete Loop.

We are competing with nobody.

This sounds like a positioning cliche, but it is literally true. Of 207 companies mapped in this space, zero operate in Latin America. Zero handle NF-e. Zero connect to Pix programmatically for agent-initiated transactions. Zero integrate with Correios, Jadlog, or Loggi for agent-initiated shipping. Zero connect to Omie, ContaAzul, or Bling for agent-initiated ERP registration.

Our positioning is: LatAm commerce infrastructure for AI agents. Not generic tools. Not US checkout. Not protocol compliance. Commerce infrastructure: the concrete, messy, regulated, domain-specific capabilities that agents need to participate in the LatAm economy.

Why LatAm is the best market for agent commerce

This is a strong opinion. We believe Latin America is not just a viable market for agent commerce infrastructure - it is the best market. Better than the US. Better than Europe. Better than Southeast Asia. Here is why.

1. High mobile, low legacy

LatAm skipped the desktop internet era. Brazil has 165 million smartphone users and only 80 million desktop internet users. WhatsApp is the primary business communication channel. Pix is the primary payment method. The entire commerce stack is mobile-first and API-first. This is the exact environment where AI agents thrive: APIs, not web forms. Messaging, not email. Real-time payments, not batch processing.

In the US, agent commerce has to integrate with legacy systems: EDI for supply chain, ACH for payments (2-3 day settlement), USPS/UPS/FedEx with APIs designed in 2005. In LatAm, the infrastructure is newer, more API-friendly, and more amenable to autonomous agent interaction.

2. Regulatory complexity = moat

Brazil has the most complex fiscal system in the world (1,501 hours/year on tax compliance, World Bank). Mexico has CFDI requirements. Argentina has Factura Electronica. Colombia has electronic invoicing mandates. Each country has its own regulatory framework, its own fiscal document types, its own tax authority APIs.

In the US, this complexity does not exist. A US company issues an invoice by emailing a PDF. A Brazilian company issues an invoice by sending an XML document to SEFAZ, receiving a signed authorization response with a 44-digit access key, storing the XML for 5 years, and making the DANFE (printed version) available to the customer. The complexity is not optional. It is legally required.

This regulatory complexity is our moat. A US competitor cannot add LatAm fiscal compliance as a feature. It requires years of domain expertise, relationships with tax authorities, and understanding of edge cases that are not documented anywhere. Every month we operate, our moat gets deeper.

3. Massive unserved SMB market

Brazil has 21 million registered businesses (IBGE). 99% are small or medium enterprises. According to Sebrae, 68% of these SMBs have been fined for fiscal errors. 78% use WhatsApp for business communication. 89% process payments via Pix. And zero of them have access to AI agent commerce infrastructure.

In the US, SMBs have access to Shopify, Stripe, Square, QuickBooks, ShipStation, and dozens of other tools that automate commerce workflows. In Brazil, SMBs use WhatsApp, spreadsheets, and manual data entry. The gap between US SMB tooling and LatAm SMB tooling is enormous, and AI agents are the fastest way to close it.

4. Real-time payment rails are agent-ready

Pix settles in 10 seconds, 24/7/365. SPEI settles in under a minute. These are real-time, API-accessible, no-fee payment rails that are perfectly designed for autonomous agent transactions. Compare this to ACH (2-3 business days, batch processing, no weekends) or wire transfers ($25 fee, manual initiation). LatAm payment infrastructure is more advanced than US payment infrastructure for agent use cases.

5. WhatsApp is the universal interface

197 million WhatsApp users in Brazil. 78% of consumers have purchased via WhatsApp. WhatsApp is the interface that agents should use to communicate with LatAm customers, not email (18% open rate), not SMS (limited to 160 characters), not a mobile app (requires download and retention). WhatsApp is already installed, already used for commerce, and already has business APIs (via Z-API, Twilio, Take Blip). An AI agent that communicates via WhatsApp is not asking the customer to adopt new behavior. It is meeting them where they already are.

What the map tells us

The category has been validated by capital, not by use. $70M+ has been invested. Most of it is pre-revenue or very early revenue. The market belief is real; the market reality is still forming. This is typical for infrastructure categories: the investment comes before the revenue because infrastructure takes time to build and adopt.

The layers are separating cleanly. Identity (Skyfire, Nekuda), vault/PCI (Basis Theory), checkout execution (Rye, Firmly), fraud (Forter). Each is becoming its own company. The stack is disaggregating. This disaggregation creates opportunities for specialized players at each layer and integration plays (like ours) that connect the layers.

The geographic gap is structural, not accidental. It will not be closed quickly by a US competitor adding a few integrations. Latin American fiscal complexity, payment rail diversity, and logistical fragmentation create a barrier to entry that compounds over time. The longer we are in production, the wider the gap becomes. Every SEFAZ rejection code we handle, every Correios API quirk we work around, every Zoop split payment edge case we solve makes our infrastructure harder to replicate.

The winner in LatAm agent commerce will not be a US company expanding south. It will be a company born in the region, building for the region, with the domain expertise that only comes from operating in the region. We believe that company is CodeSpar.

Predictions: the next 36 months

Based on the patterns in this landscape, the funding trajectories, the regulatory environment, and the technology maturity curves, we make seven specific, falsifiable predictions for the agentic commerce space through 2028.

Strategic advice: if you are building in this space

We will close with unsolicited advice for founders, investors, and product leaders evaluating the agentic commerce landscape. This comes from mapping 207+ companies, analyzing $2.4B+ in capital deployment, and building in the space ourselves.

For founders

Pick a layer and a geography. Do not try to be horizontal. The US market is crowded in every layer. Generic, US-focused, single-layer companies will be commoditized or acquired within 24 months. The winning strategy for new entrants is geographic specialization (LatAm, Southeast Asia, MENA) or deep vertical specialization within a layer (healthcare agent payments, real estate agent transactions).

Build on MCP. Do not bet on a proprietary protocol. The protocol wars are effectively over at the developer layer. MCP has the ecosystem. Build your commerce tools as MCP servers. If ACP or UCP wins in specific domains, you can add support later. But your developer adoption engine should be MCP-native from day one.

Capture per-transaction revenue, not just SaaS fees. The companies that will reach $100M+ ARR fastest are the ones with transaction-based revenue that scales with GMV. SaaS fees plateau. Transaction fees compound. If your product touches the money flow, capture a percentage of every transaction. This is why Stripe is worth $95B and protocol companies raise $5M seed rounds.

If you are building in LatAm: The window is open but will not stay open forever. Right now, zero companies serve this market. In 18 months, there will be 3-5. In 36 months, the early movers will be entrenched. The infrastructure play is the most defensible. The application play is faster to revenue but less defensible. Domain knowledge is the real moat, not technology. Hire domain experts, not just engineers.

For investors

The geographic arbitrage play is wide open. Of 207+ companies, zero operate in LatAm, zero in Southeast Asia (680M population), zero in Africa (1.4B population). The US has $2.4B in funding and 150+ companies. LatAm has $340B in e-commerce, $2.1T in digital payments, and zero funded agent commerce infrastructure companies. The opportunity-to-competition ratio is extreme.

Bet on execution layers, not protocol layers. Protocols will be free and open. Execution layers capture revenue. The best risk-adjusted investments are companies that own the transaction execution point in specific markets. Look at the gap matrix: the most funded categories (payments & identity) are the most crowded. The least funded capabilities (fiscal compliance, LatAm logistics, ERP integration) are the most defensible. Fund the capabilities that are hard to build, not the capabilities that have the most buzz.

Cross-layer companies in specific geographies are the category kings. Stripe became a $95B company by integrating payments, identity, fraud, and merchant tools across a single platform. The equivalent play in agent commerce is integrating protocol tooling, payments, checkout, and merchant enablement in a specific geography. This is the CodeSpar thesis.

For enterprise leaders

Start with the Complete Loop, not with “we need an AI strategy.” Start with “we have 200 orders per day and the post-sale workflow takes 14 hours per week.” Agent commerce is not a technology decision. It is an operations decision. The technology is ready. The question is whether your operations team is ready to let an agent handle the deterministic 95% so humans can focus on the exceptional 5%.

If you operate in LatAm, do not wait for US companies to add support. The structural barriers to entry mean US agent commerce companies will take 2-3 years to build adequate LatAm coverage, if they do at all. By then, the LatAm-native infrastructure will be entrenched. Evaluate LatAm-specific solutions now.