Y Combinator-backed Rulebase wants to be the AI coworker for fintech

Y Combinator-alum Rulebase is betting that the adjacent question of automation successful fiscal services won’t beryllium astir flashy AI interfaces, but the unglamorous back-office tasks similar compliance.

The startup, founded by Gideon Ebose and Chidi Williams, 2 Nigerian engineers who met successful London, conscionable raised a $2.1 cardinal pre-seed circular led by Bowery Capital, with information from Y Combinator, Commerce Ventures, Transpose Platform VC, alongside respective angels.

Financial services firms walk tremendous amounts of effort connected enactment tickets, resolving disputes, ensuring prime assurance, and regulatory compliance. Rulebase’s software, which it calls an cause coworker, replaces overmuch of the manual grunt enactment successful these tasks. Its AI cause tin measure lawsuit interactions, emblem regulatory risks, and trigger the close follow-ups crossed tools similar Zendesk, Jira, and Slack without losing the human-in-the-loop oversight that fiscal firms demand.

“Our ‘Coworker’ instrumentality integrates crossed platforms and collaborates with quality agents and back-office teams to afloat negociate the quality lifecycle portion redeeming time, reducing errors, and maintaining compliance,” said CTO Williams. Currently, the year-old startup is already deployed astatine customers similar U.S. concern banking level Rho and an unnamed Fortune 50 fiscal institution.

Rulebase wasn’t the founders’ archetypal swing. Ebose, a erstwhile merchandise pb astatine Microsoft and Williams, a erstwhile backend technologist astatine Goldman Sachs, built respective products unneurotic similar an AI lawsuit feedback instrumentality earlier yet settling connected Rulebase. The thought came astir aft seeing however inefficient back-office operations were successful tiny and ample fiscal institutions, particularly erstwhile it came to regulatory workflows.

The startup presently focuses connected workflows triggered by lawsuit work interactions, with its archetypal wedge astir prime assurance. QA analysts successful accepted fiscal institutions typically manually reappraisal 3–5% of enactment interactions to guarantee reps travel compliance protocols. 

Rulebase present evaluates 100% of specified interactions, cutting costs by up to 70%, the founders say. In the lawsuit of Rho, for instance, Rulebase has helped chopped escalations by up to 30%.

“We automate workflows that commencement with a lawsuit interaction, areas we’re already large astatine handling end-to-end,” CEO Ebose said successful an interrogation with TechCrunch. “While overmuch of that is QA, compliance, and disputes tied to lawsuit calls and messages, semipermanent our extremity is to instrumentality connected arsenic galore manual back-office tasks arsenic imaginable by pulling these fragmented steps and tabs into 1 coordinated workflow.”

The caller backing volition assistance it treble down connected engineering and yet adhd caller features to AI Coworker similar fraud investigation, audit preparation, and regulatory reporting.

Rulebase is focused connected fiscal services for present due to the fact that automation demands precision. “You request to recognize MasterCard’s rules, CFPB timelines. That extent of domain cognition is our moat,” Ebose said.

They institution is targeting concern banks, neobanks, and paper issuers crossed Africa, Europe, and the U.S. But the roadmap could yet see adjacent verticals similar insurance, wherever akin workflows exist.

Screenshot

Revenue is increasing fast, with “double-digit” month-over-month maturation since joining Y Combinator’s Fall 2024 batch, the founders say. Rulebase’s concern exemplary is usage-based, charging per enactment reviewed oregon workflow automated.

As 1 of the fewer African founders to get into YC gathering AI tools, Ebose and Williams’ proposal to founders trying to get accepted into the planetary accelerator is to deliberation globally from time one.

“We’re successful a infinitesimal wherever tiny teams tin present much value, much quickly, than ever before, truthful limiting yourself to ‘X for Y’ oregon a constrictive vertical feels similar a missed opportunity,” Williams remarked. “With AI, it feels evident that you person to spell aft thing massive. Anything little than the astir ambitious mentation of your thought apt won’t chopped it,” said Williams, who earlier Rulebase built Buzz, an aboriginal open-source speech-to-text instrumentality with implicit 300,000 downloads and implicit 12,000 GitHub stars.