Rulebase, a Y Combinator alum, is betting that the next wave of automation in financial services will focus on unglamorous back-office tasks like compliance rather than flashy AI interfaces. The startup was founded by Gideon Ebose and Chidi Williams, two Nigerian engineers who met in London. It just raised a $2.1 million pre-seed round led by Bowery Capital, with participation from Y Combinator, Commerce Ventures, Transpose Platform VC, and several angel investors.
Financial services firms spend enormous effort on support tickets, resolving disputes, ensuring quality assurance, and regulatory compliance. Rulebase’s software, which it calls an agent coworker, replaces much of the manual work in these tasks. Its AI agent can evaluate customer interactions, flag regulatory risks, and trigger the right follow-ups across tools like Zendesk, Jira, and Slack, all while maintaining the human-in-the-loop oversight that financial firms require.
The company’s CTO, Chidi Williams, stated that their Coworker tool integrates across platforms and collaborates with human agents and back-office teams to fully manage the dispute lifecycle while saving time, reducing errors, and maintaining compliance. Currently, the year-old startup is already deployed at customers like U.S. business banking platform Rho and an unnamed Fortune 50 financial institution.
This venture was not the founders’ first attempt. Ebose, a former product lead at Microsoft, and Williams, a former backend engineer at Goldman Sachs, built several products together, including an AI customer feedback tool, before settling on Rulebase. The idea emerged after they observed how inefficient back-office operations were in both small and large financial institutions, particularly regarding regulatory workflows.
The startup currently focuses on workflows triggered by customer service interactions, with its initial entry point being quality assurance. In traditional financial institutions, QA analysts typically manually review only 3 to 5 percent of support interactions to ensure representatives follow compliance protocols. Rulebase now evaluates 100 percent of such interactions, which the founders claim cuts costs by up to 70 percent. For example, at Rho, Rulebase has helped reduce escalations by up to 30 percent.
CEO Gideon Ebose explained in an interview that they automate workflows that start with a customer interaction, areas they are already great at handling end-to-end. While much of that is QA, compliance, and disputes tied to customer calls and messages, their long-term goal is to take on as many manual back-office tasks as possible by pulling fragmented steps and tabs into one coordinated workflow.
The new funding will help the company double down on engineering and eventually add new features to its AI Coworker, such as fraud investigation, audit preparation, and regulatory reporting. Rulebase is focused on financial services for now because automation in this field demands precision. Ebose noted that you need to understand MasterCard’s rules and CFPB timelines, and that depth of domain knowledge is their moat.
The company is targeting business banks, neobanks, and card issuers across Africa, Europe, and the U.S. However, the roadmap could eventually include adjacent verticals like insurance, where similar workflows exist.
Revenue is growing fast, with the founders reporting double-digit month-over-month growth since joining Y Combinator’s Fall 2024 batch. Rulebase operates on a usage-based business model, charging per interaction reviewed or workflow automated.
As one of the few African founder teams to get into Y Combinator while building AI tools, Ebose and Williams advise other founders trying to get into the global accelerator to think globally from day one. Williams remarked that we are in a moment where small teams can deliver more value more quickly than ever before, so limiting yourself to a narrow vertical feels like a missed opportunity. With AI, it feels obvious that you have to go after something massive. He stated that anything less than the most ambitious version of your idea likely will not cut it. Before Rulebase, Williams built Buzz, an early open-source speech-to-text tool with over 300,000 downloads and over 12,000 GitHub stars.