Fintechs have taken advantage of the boom in embedded finance to offer a range of banking and payment solutions, but the tax industry has been relatively slow to make progress. This reluctance may be due to the tax industry’s complex ecosystem and regulatory web, where legislative differences at multiple levels force tax software providers to walk on eggshells.

Companies like Credit Karma and Intuit offer bundled tax services, but other fintechs are pursuing alternative strategies – such as partnerships or developing their own technologies – to establish their presence.

Lili, for example, has had success in combining banking, payments, invoicing and tax management into a single app for small and medium-sized businesses (SMEs).

Another player in the broader market is april, a fintech startup that partners with banks, fintech companies and payroll providers like Gusto to provide them with embedded tax solutions. april’s tools help users file, estimate and optimize their taxes directly in the financial apps they already use. Rather than go the tax technology partnership route, april decided to build its own infrastructure from scratch.

We explain how April built its platform and manages its infrastructure to offer e-filing capabilities in today’s market, the challenges involved, and whether the market today sees standalone embedded tax offerings as viable.

Platform building: The alchemy of taxes and technology

The fintech company was initially founded in 2018 under the name Limepay as a white-label BNPL provider and renamed in April last year to highlight its focus on its core offering of AI-driven tax software.

To make its vision of integrating tax and technology a reality, april needed to build and invest in a competent team and solid infrastructure.

Building the team

To develop the platform, April hired engineers and paired them with tax experts and a virtual assistant for personalized tax coding.

Building a team dedicated to tax was critical because tax is an integral part of april’s operations, giving the company a greater reach across the business. “Everything we want to do will be tax-weighted at its core,” said Christina Taylor, april’s vice president of tax development and execution.

To integrate tax expertise into all aspects of her operations, support and development, Taylor leverages her nearly twenty years of tax, accounting and business operations experience to mentor her team of tax analysts to drive better results. Her previous roles include leading teams at Credit Karma Tax and Cash App Taxes.

“I make sure they have the guidance and tools they need to do that, and I oversee the processes to make sure we’re all aligned as tax analysts and engineers,” she said. “When I joined April a year ago, I brought in top-notch analysts and veterans from the tax industry because I wanted to build a team with the skills needed to take April nationally.”

Development of the internal technical infrastructure

U.S. tax law is a labyrinthine challenge that makes developing effective digital and mobile solutions difficult. April’s strategy to address this challenge involves a two-pronged approach: addressing technology and tax separately before combining them into a unified product. This hybrid model brings together technical and tax expertise, combining digital advances with human knowledge.

To connect tax experts with engineers and keep pace with evolving federal and state tax policies, april relies on its proprietary AI system, Tax-to-Code. This system uses specially trained Large Language Models (LLMs) to analyze tax documents (Technical Product Requirements Documents) and generates code suggestions to define data entries, logic, and relationships for precise tax calculations. These suggestions are then reviewed by april’s tax engineers, who focus on improving the accuracy of the LLMs with each technical PRD they process.

However, this is easier said than done.

“One of the challenges of developing a tax product is that the laws are constantly changing and you have to not only keep track of the changes, but also stay one step ahead of them,” Taylor noted.

This means that the algorithms must be constantly optimized to adapt them precisely to the requirements.

The race against time when programming: A tax product can include millions of lines of code to achieve nationwide coverage. The federal tax code alone has grown to over 70,000 pages and more than 10 million words, with annual changes at the state and federal levels.

Taylor notes that developing a single tax filing product is like developing 43 different products for federal and state taxes. “I like to say that all states are like their own sovereign entities when it comes to tax issues – they all have different ideas, techniques, tax laws, platforms and technologies,” she said.

Analyzing each state’s tax laws is a daunting undertaking in itself, and coding it into a working e-file product adds to the complexity. The need for thorough testing and timely updates further exacerbates the challenge.

Although navigating the ever-changing tax law is a complex task, this method gives April a competitive advantage over traditional players.

“The biggest difference between the way we develop our tax products and the way the previous vendor does is that our engineers primarily build internal tools rather than the actual logic of our product,” Taylor added.

By sticking to this approach, Taylor says April has achieved coverage of nearly 85% of all U.S. tax scenarios within three years and has established nationwide electronic filing for both state and federal taxes.

The right dose of AI: AI is widely implemented in April’s backend operations and includes systems such as the generative AI process Tax-to-Code and aprilOS, which facilitates repetitive tasks.

“Using AI, we can essentially increase the speed at which we code tax laws tenfold,” Taylor explained.

Also on the front end, an AI-powered chatbot uses a combination of AI models and input from April’s support team to provide users with algorithmically generated answers.

Even though the company relies heavily on AI, it does not blindly trust this technology. The fintech company is aware of the limitations of AI models and has all AI-generated results checked by its engineers.

“Taxes are complex, sensitive and highly regulated, so we see no benefit in AI-washing here,” Taylor explained. “We have carefully integrated AI into our processes.”

Bundled vs. standalone embedded tax products – the better approach today?

Taylor points out that the tax industry has come a long way since the advent of fintech technology, but it’s a long way from 10 to 15 years ago when professionals like her struggled to communicate the value of tax services.

Today, customers have the option to offer tax services for free, for a fee, or as part of a package. However, there is a growing trend among fintechs to bundle tax services.

Taylor also sees standalone embedded control products as a promising opportunity. “That’s why I joined April – that and to use AI to make things better and smarter,” she added.

However, she argues that taxes should not be isolated as a one-off service offering.

“Taxes shouldn’t be an isolated product – they’re tied to so many aspects of our financial lives – income, investments, spending,” she said. “I believe making tax data accessible will enable more Americans to make more informed financial decisions and file their taxes faster than ever before when the time comes.”

Tax data can also be a resource for fintechs to offer personalized products and services. When a consumer’s tax return is electronically filed and accepted, fintech apps can use the verified tax data to offer their users customized services, such as mortgage loans or car financing rates.

“We believe taxes are a broader means to an end – a smarter consumer financial technology ecosystem,” Taylor noted.