N3XT, a blockchain-based bank focused on programmable B2B payments, has launched N3XT MCP, an implementation of the open Model Context Protocol standard designed to connect corporate AI environments directly to live banking data. The product went live on 28 July 2026.
The architecture allows AI agents operating inside a company’s own infrastructure to query balances, draft payments, generate reports and trigger escalation workflows without requiring manual data exports or switches between portals. N3XT says the integration preserves existing compliance rules and user permission structures, including maker-checker controls, and keeps queries within the corporate firewall rather than routing data through external AI services.
What MCP does in practice
The Model Context Protocol is an open standard, not proprietary to N3XT, that defines how AI agents can request context from external systems in a structured, permissioned way. N3XT’s implementation sits on top of its existing API layer and extends it to agentic use cases: rather than a developer calling an API programmatically, an AI assistant can invoke the same capabilities in response to a natural-language instruction from a treasury operator or finance analyst.
The practical workflow N3XT describes includes automated discrepancy alerts, smart escalation and scheduled report generation. The governed write capability is the more consequential feature: it means an AI agent can draft a payment instruction, subject to the same approval chain a human user would face, rather than only reading data passively.
Aurélien Bonnel, founder and chief technology officer of N3XT, said the launch is intended to standardise how AI agents access live bank data. “We are standardizing how AI agents access live bank data and enabling corporate clients to work within their own environment with data they can trust,” he said.
Market context and competitive landscape
The treasury and corporate banking technology market is seeing a wave of MCP-adjacent integrations as enterprise software vendors race to make their data layers compatible with large language model tooling. Several banking-as-a-service providers and core banking platform vendors are pursuing similar connectivity, though most have framed it as API enhancement rather than a dedicated agentic protocol layer.
N3XT’s positioning as a narrow bank built on blockchain, settling in on-chain US dollars, places it in a narrower competitive set than a conventional treasury management system vendor. Its direct competitors are more likely to be programmable payment infrastructure providers and stablecoin-adjacent corporate banking platforms than incumbents. The durability of the advantage depends on whether the MCP interface attracts enterprise treasury clients who prefer keeping AI workloads inside their own environment, a genuine concern given the sensitivity of real-time payment data.
From a regulatory standpoint, agentic write access to payment rails sits in an area that regulators are beginning to scrutinise. In the UK, the FCA‘s ongoing AI and digital innovation work touches on automated decision-making in financial services. In the US, the OCC’s guidance on bank technology third-party risk is relevant to any corporate client deploying an AI agent with payment-drafting capability. N3XT’s emphasis on firewall containment and existing permission structures is likely a direct response to these concerns, though clients will need to satisfy themselves that the maker-checker workflow survives agentic automation without introducing new operational risk.
The next milestones to watch are named enterprise client deployments, independent security audits of the MCP layer, and whether N3XT progresses toward a US bank charter or continues to operate under its current licensing structure.