Omise AI
AI-powered efficiency
Back
Products
For Merchants
For Platforms and Banks
Professional Services
Resources
Help center
Despite all innovation the Fintech industry has invested in, there’s one area that remains notoriously slow: merchant onboarding.
Payment providers prioritise safety and regulatory compliance, so it makes perfect sense that thorough checks are required. However, sometimes it takes more than a week to onboard a merchant, and during this process, we lose customers. For example, in 2024, a study of over 150 C-level executives from corporate, institutional, and commercial banks in Singapore revealed that almost 90% have lost clients over the past year due to delays and inefficiencies in onboarding. Translated into monetary terms, the loss is unimaginable.
Time consumption isn’t the only issue. Manual merchant onboarding also carries the risk of human error and inconsistency. Across the banking, fintech, and crypto industries, there have been numerous cases where companies faced major penalties and reputational damage due to failures in KYC and AML controls.
What, then, is our vision to solve these problems? Fully autonomous onboarding with MCP.
AI assistants today can do a lot. They answer questions, summarize information, and even write code. But their potential goes far beyond that. To become more advanced, accurate, and relevant, they need better access to the tools and data that drive real-world tasks. What often stands in the way are fragmented systems: data trapped in legacy infrastructure and tools that require custom-built integrations.
This is where MCP (Model Context Protocol) comes in.
MCP answers the question: what if every data source, tool, and service had a standardized way for AI agents to connect? Developed by Anthropic, the company behind Claude AI, MCP is an open-source standard that allows AI agents to connect to any data, tool, or service through a single, unified protocol. As Anthropic describes it, it's “like USB-C for AI”: a universal connector that lets agents plug into any compatible server, eliminating the need to build custom integrations for every tool.
Imagine giving AI a single prompt, and it seamlessly pulls information from multiple content repositories and chains together various tools to work in a goal-driven, intelligent way — all to get the task done for us. That’s exactly what we’re aiming for in the merchant onboarding process.
In our case, the goal is to automate the entire onboarding journey. Our envisioned workflow consists of the following key components:
By adopting the MCP approach, onboarding can shrink from days to minutes, while the system scales seamlessly across regions with different KYC and AML requirements, making fully autonomous, error-resistant onboarding a reality for every merchant. That’s the picture Omise is aiming for.
Just like it does with autonomous onboarding, MCP is what turns Agentic Commerce from a smart idea into a seamless experience. Think of it as a universal connector that lets your AI shopping assistant talk to every tool it needs: storage systems, product databases, price comparison engines, checkout platforms, and payment gateways, without the friction of custom connections.
Because everything speaks the same “language” through MCP, your AI assistant can move smoothly from one step to the next: finding the perfect item, checking if it’s in stock, comparing prices, applying promotions, and completing a secure checkout. It’s all done in the background.
The results are a smarter and more intuitive shopping experience that’s tailored to customers, and for marketplaces or e-commerce websites with millions of items? Less decision paralysis and less drop-offs at checkout.
Agentic Commerce is the future. From discovery to checkout in a single conversation, to buying and sending gifts before a deadline, to automatically purchasing when prices drop, and more. That is why so many are investing in building shopping assistants, each one smarter than the last. With MCP at its core, the future is already taking shape.
In our next article, we will share our journey in building Omise’s Agentic Commerce, explore smart technologies that make it possible, and reveal our first demo for a real client. Stay tuned!
Resources
Fintech News Singapore. (2025, February 5). Slow onboarding drives client losses in Singapore’s banks. Fintech Singapore. https://fintechnews.sg/107398/digital-banking-news-singapore/slow-onboarding-drives-client-losses-in-singapores-banks/
Lisowski, E. (2025, August 4). MCP explained: The new standard connecting AI to everything. Medium. https://medium.com/@elisowski/mcp-explained-the-new-standard-connecting-ai-to-everything-79c5a1c98288
Introducing the model Context Protocol. (n.d.). https://www.anthropic.com/news/model-context-protocol
