Sokin has launched a Model Context Protocol (MCP) connector that links its licensed payment infrastructure directly to AI tools including ChatGPT, Claude, Gemini, and Perplexity. The connector, announced September 14, 2026, lets finance teams prepare cross-border payments, retrieve FX quotes, and manage balances from inside a chat interface. For operations teams, it cuts out the manual data entry that typically eats hours of accounts payable work.
What the connector does
The MCP connector moves beyond read-only AI integrations. An authorized user's AI agent can create new beneficiaries, retrieve real-time foreign exchange quotes between supported Sokin wallets, and prepare currency exchanges. It can also find existing beneficiaries or validate new beneficiary details against specific payment route requirements.
In each case, the AI tool assembles the full instruction and presents the details to the user. Execution only happens after a human confirms the action, and the instruction follows the customer's pre-configured approval rules and internal governance flows. The AI prepares the work; the human still signs off.
During testing, Sokin reported the connector processed 10 multi-currency invoices and treasury moves in under 10 minutes. That is work an accounts payable team would normally key in by hand, often taking hours.
The infrastructure underneath
The interface is new, but the settlement runs on Sokin's existing licensed rails. The company holds licenses across 36 countries and moves money to more than 170 countries in over 70 currencies. A prepared instruction settles on infrastructure Sokin owns and is licensed to operate.
"A read-only connector, which is what most MCPs in this space are right now, is a search bar with more personality," said Pete Daunton, Chief Strategy Officer at Sokin. "Our MCP reaches the tools that create a beneficiary, quote an exchange rate and prepare a payment. The ability to give an agent real work is what is going to save our customers a lot of time and what they are going to love most about this."
Sokin customers already access the platform through a portal or embedded APIs. The connector adds a third route, through compatible AI and Large Language Models. For teams evaluating MCP Courses or building skills in AI for Finance, the shift from information retrieval to action is the core development here.
What changes for the industry
For years, AI in finance has been limited to dashboards and chatbots that answer questions but hand the work back to humans. The Sokin connector is an execution-based integration: the AI agent prepares real payment instructions that run on licensed rails after approval.
This puts pressure on legacy banking institutions whose closed systems do not fit the open-standard nature of LLM connectors. The remaining challenge is not technical capability but regulatory trust. Can the industry maintain AML and fraud standards when the primary interface for a payment is a third-party AI agent?
Why this matters for operations teams
Operations teams spend a disproportionate amount of time on repetitive financial tasks: keying in invoices, checking beneficiary details, chasing payment statuses. A connector that lets an AI agent handle preparation and present a complete instruction for approval removes hours of manual work per week. The approval step stays with humans, which means existing governance controls remain intact. The practical takeaway: evaluate whether your current payment infrastructure can support agentic workflows, because the tools are arriving faster than most back-office processes can absorb them.
Your membership also unlocks:
