The Tax Habit with Ben Borodach

The American tax system handles trillions of dollars annually, spans dozens of federal and state jurisdictions, and touches virtually every major financial decision a household makes. 

As shown in the graph above, the US Department of the Treasury has recorded $2.48tn in taxes in FY 2026 to date, approximately 55% of which is personal income tax. Despite this colossal figure, explains Ben Borodach, Co-founder and CEO of April, the technology supporting tax filing remains catastrophically outdated. The software that most Americans use to navigate their tax obligations was designed decades ago.

Borodach began building April’s “tax engine” five years ago, before large language models (LLMs) were a mainstream concept. While artificial intelligence (AI) is integrated into April’s services, he admits that more comprehensive change is needed to support taxpayers throughout the year. In the latest edition of CMC Aureon Sessions, he walks listeners through the technological, behavioural and structural barriers that stand between the current state of tax in America and something meaningfully better.

Fragmentation and hyper-personalisation

To explain the opportunity in tax technology, Borodach points to the trajectory of the broader financial services industry over the past 15 years. The emergence of banking-as-a-service, the explosion of consumer fintech apps, and the progressive democratisation of investment and payments products have put more financial capability in more hands than at any point in history. 

This proliferation has come at a cost, however.

“People’s financial lives ended up all over the internet,” Borodach points out. “I have a banker; I have this financial advisor I inherited. I use this money app. I’m buying crypto over here. I’ve got this other thing to manage my small business. My kids are on this app, my spouse is on that app. And suddenly you’re like, how am I making financial decisions?”

This fragmentation creates an acute problem for financial institutions competing for relevance and wallet share. The ease with which consumers can now open accounts has eroded the structural advantages that larger, more established players once enjoyed. As Borodach notes, the competitive battleground has shifted from product access to engagement and trust. “Opening an account used to be hard. Now it’s pretty easy. The hard part is everybody is fighting for account primacy, share of wallet, the customer really engaging with your platform.”

The strategic response emerging across the industry is hyper-personalisation: the ambition to move from blunt, one-size-fits-all product experiences towards services that respond to the specific circumstances, life stage and financial profile of each individual customer. Borodach illustrates the gap between ambition and reality with an anecdote about a major credit card company that claimed deep knowledge of its customers, yet responded to the birth of his child by sending discount offers at Best Buy, and to a career promotion with a coupon for a random restaurant. “How was that an effective form of communication based on my needs? You clearly didn’t really know me, so I ended up somewhere else.”

The insight that underpins April’s strategy is that tax is uniquely positioned to close this personalisation gap. A tax return provides a comprehensive portrait of a person’s financial year, depicting income changes, major life events, investment activity and emerging needs in a way that no other data source can match. For the financial institutions that can access and act on that information, the opportunity to deepen customer relationships is substantial.

“Financial services companies make their business by being available at that moment,” Borodach explains. “That’s what our tagline at April is: be there for the moments that matter.”

The limits of AI hype

No serious discussion of financial technology in 2026 can avoid the question of AI. The past few years have seen a wave of products that claim AI capabilities – with varying degrees of substance behind those claims. Borodach’s perspective on this is more nuanced than most, because April’s bet on AI predates the mainstream adoption of LLMs and was made for specific, well-considered technical reasons.

“AI can both be very real and be overhyped a little bit,” he admits. “That’s our assessment of LLMs.”

While April’s tax engine is deterministic – meaning it produces consistent, auditable outputs from defined inputs – Borodach is actively sceptical of the idea that LLMs, given their black box nature, should be trusted to perform tax calculations.

“We believe that the engine needs to be deterministic … [but] we do not believe, for example, that tax calculations should be done by probabilistic, unexplainable, unauditable LLMs… And I think venture capitalists that are betting on LLMs being calculation engines in different parts of the market are going to be sorely disappointed.”

This is a significant call in a market where AI-powered tax products are proliferating. Borodach’s argument is not that LLMs have no role in tax technology – April uses them extensively – but that they should be deployed where their strengths align with the task at hand, rather than applied wholesale to problems requiring exactitude.

In practice, April has found productive applications for AI across several dimensions of its business. Optical character recognition capabilities – used to extract data from tax documents – have allowed April’s platform to move from two supported tax forms to 15, with Borodach forecasting it will be able to handle 25 next year, powered in part by improvements in models and leveraging April’s proprietary data. 

In some instances, AI offers a competitive advantage over a human worker. “We use it in customer support,” he explains. “In our self-prep product, a very high percentage of customer contacts have been resolved using an AI agent. In a lot of cases, it’s faster and more reliable than a human.” 

However, he places central importance on human capital. “We are still growing our human team and we have human support for escalations or problem sets that are not suited for the AI.”

In Borodach’s view, the more interesting frontier for AI lies in tax planning, simulation and scenario analysis. Services like Holistiplan and Altruist’s Hazel product have brought LLM-powered insights to financial advisors, but Borodach believes April’s differentiator will lie in moving beyond eligibility flags to precise, quantified tax impact analysis.

“It’s one thing to say, Ben might be eligible to do something. It’s another thing to say, here’s the precise tax impact that is going to happen. That’s the world I foresee us differentiating our services and our IP around.”

His broader framing of AI’s role draws an analogy with cloud computing – another technology that was initially met with scepticism, then adopted rapidly by startups before reshaping incumbents’ strategies. The transformative applications, in his view, are those where AI is genuinely load-bearing infrastructure, not a marketing layer applied to an existing product.

“I see the AI revolution much more similar to cloud. In the beginning, people were scared of the cloud. They didn’t want to lose control. But startups went there very quickly and realised, ‘oh my God, now I’m liberated.’”

Changing taxpayer behaviour

In the end, addressing consumer behaviours around taxes may be more challenging than the adoption of new technologies. Borodach notes that the way that most Americans relate to their taxes is a deeply ingrained pattern of behaviour that technology alone cannot dislodge.

“When I look at tax, the problems are because people only touch it a week or two a year and they leave it for the last minute,” he says. “Imagine if you were trying to get healthy and you were like, I have my physical in two weeks, time to get on the treadmill … You’re not really going to get the results you want.”

The analogy is instructive. Just as health outcomes depend on sustained daily habits, financial outcomes depend on ongoing tax awareness that stretches beyond tax season. Borodach’s argument is that the structural conditions for this shift are now in place: the embedded financial platforms that Americans use daily are, for the first time, capable of surfacing tax-relevant information and nudges in the right context, at the right moment. “It’s like nutrition, physical activity, savings. You can’t just do it once in a while when you want. It’s actually a habit and part of your vernacular you have to get comfortable with.”

The specific pain points that compound when tax is treated as a once-a-year event are numerous and financially significant. Many Americans set their W-4 withholding once, forget about it and discover at filing time that they owe more than they expected – or end up delivering an interest-free loan to the federal government through an unnecessary refund. As a result of misinformation about taxes, many people “don’t set enough money aside at the end of the year to pay the IRS or they end up with this large tax refund, which is kind of a waste. They could have invested it or used it for some other purpose in their life.”

Borodach is clear-eyed about the limits of what technology can accomplish without a parallel shift in how people think about tax. He compares the potential shift to the impact of Apple’s [AAPL] iPhone, in that it fundamentally changed what people believed a mobile device should do. “I think it doesn’t matter what the technology does,” he says. “Until we actually change the way that people think about this, we’re not going to actually effect the future that the American public wants and deserves.”