For the last 20 years Dr Beibei Flynn has worked in large-scale AI systems research, development and deployment across the financial and tech industries.
She has a PhD in machine learning and is a qualified chartered tax adviser, and while working for Accenture, Flynn was the driving force behind three machine learning projects with the Irish Revenue.
Earlier this year, she gathered all of her experience together to set up Leapifai, an AI-driven platform that replaces time-consuming manual research with instant and verifiable tax advisory guidance for accountants and tax professionals.
“While working on the machine learning projects, I discovered that true innovation in technology demands more than just technical expertise. It requires a deep grasp of the business itself,” Flynn says. “This led me to pursue the so-called gold standard in Irish tax qualifications – chartered tax adviser – which turned into nearly seven years of relentless study, sacrificing weekends and holidays and overcoming multiple exam setbacks along the way.”
But while the path to the qualification was difficult, it opened Flynn’s eyes to the challenges faced by accountants who are under intense pressure to meet a suite of reporting deadlines. “To ensure precision, they need quick and reliable guidance when it comes to tax, and Leapifai was born to help them avoid the challenges I experienced along the way,” says Flynn, who recently participated in an AI ecosystem accelerator run by NovaUCD and CeADAR (Ireland’s national centre for AI).
The starting point for Leapifai was building its knowledge/tax research database, which meant consolidating all of the information related to tax law, appeals to the tax commission, Irish Revenue guidance on tax matters and tax-related legal cases in one place. This amounts to thousands of documents, which practitioners can now access at the click of a button.
Practitioners can interrogate the system to get clear answers to tax queries across a broad range of headings, including personal and corporation tax, VAT, stamp duty and CGT. “Tax is very broad and clients differ in size and complexity. In the past, practitioners would have waded through pages and pages of text to find the paragraph that was relevant to their query,” Flynn says. “Leapifai allows them to search the system for instant answers and breaks down complex issues into step-by-step logic, mimicking a tax adviser’s thought process, while providing referenced answers backed by law and official guidance.”
The system also allows practitioners to centrally manage and track client queries, record query history, copy and paste tax insights into client reports and stay updated on the latest tax changes.
“The success of Irish business requires significant interaction with the global economy and this imposes inherent requirements to integrate with global tax codes. When a product is designed and built to be global, it naturally positions the company to pursue international growth. Globalisation is in the DNA of Leapifai,” says Flynn, who adds that one of the biggest challenges she faced with Leapifai was building a real-time system capable of scaling to international level.
The MPV cost in the region of €300,000 to develop and the company is actively seeking investment to bring its seed round up to €750,000. The revenue model is SaaS. The system has been soft launched and will go fully commercial in the third quarter.
Flynn sees the company’s other big challenge as changing mindsets because the accountancy sector has “not historically been quick to embrace new technologies. However, this gives us a unique opportunity to lead change and help clients gain the competitive edge with our innovation,” she says.
“Accountancy is often seen as a necessary evil, especially by most strategic business managers who view it as adding little business value beyond regulatory compliance. Driving efficiency is crucial for maintaining margins for accountancy firms and balancing efficiency and accuracy are not just about compliance, they are also key commercial challenges,” Flynn says.
