Building Settla: An AI-Powered Payment & Settlement Automation Platform
This is a submission for the MLH x DEV Writing Challenge What I Built For the Paytm AI Hackathon , I built Settla , an AI-powered payment and settlement automation platform designed to simplify payment operations, refunds, vendor management, and workflow automation. The idea came from a simple observation: payment operations can become complicated when you have to deal with multiple transactions,…
This account is about a payment and settlement automation platform called Settla, which was created for the MLH x DEV Writing Challenge during the Paytm AI Hackathon. Settla is an AI-powered platform that simplifies payment operations, refunds, vendor management, and workflow automation.
The primary goal was to create a platform where AI is not just a chatbot but becomes an integral part of the workflow. Therefore, Settla was built as a full-stack application with a focus on automation and extensibility. The platform's architecture consists of different layers such as the user interface, backend API, AI/Model Context Protocol (MCP) layer, payment services, and settlement/refund workflows.
During the development process, the author emphasized the importance of testing, as payment-related workflows require predictable behavior. The project includes 135 backend tests, 9 frontend unit tests, and 13 browser E2E tests. The author learned that AI is just one part of an AI-powered product, and building everything around it is crucial for its success.
The author worked with n8n and Model Context Protocol (MCP) to connect AI systems with external tools. n8n allows workflows to be composed visually and connected to different services, while MCP allows AI models to interact with tools that expose real application functionality. This separation between intelligence, tools, and business logic proved valuable in building Settla.
The author participated in the Paytm AI Hackathon but did not win. However, the experience was still valuable, as it brought together developers, AI enthusiasts, and builders to discuss various ideas and challenges. The hackathon environment forced the author to think carefully about what to build, test, and demonstrate within a limited timeframe.
In conclusion, Settla began as a hackathon project but evolved into an experiment in how AI can interact with real systems, rather than simply generating responses. The author learned valuable lessons about building AI-powered systems, explored MCP and workflow automation, and gained insight into the challenges of presenting a product effectively.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
