What to learn from Intuit and H&R Block using AI in their Tax Prep Platforms

If you’ve done your tax prep online this year you have likely seen both Intuit and H&R Block’s initial chat and AI platforms as part of your experience. They both present lessons for companies seeking to adopt AI as part of their business that every leadership team should take to heart.

First H&R Block, which has a great story on the Microsoft site about their use of Open AI. The experience is a Q&A experience that is side-loaded with your tax prep activities that answers common questions. Here’s a quick look at it.

The experience is branded, looks great, and seems to answer questions fairly intuitively. You can see the +1 -1 buttons to gather feedback, as well as the references of where the information is coming from.

The second example is Intuit, which is much more tightly scoped, but closer to the rules engine in its execution. It isn’t as good at answering general questions, but is very well integrated into the actual workflow.

So, what to learn from each of these experiences with each major platform? There are a few elements that each has done well.

Number 1 – The platforms are true to their mission. Each of the platforms understand the job-to-be-done for their customers. In this case “do your taxes”, but even more, they understand it is to make the process as painless and beneficial as possible. Using an online service was already incredibly disruptive to the “have someone do your taxes” service, but now it is even easier and more accessible to a greater quantity of individuals. It is nearly painless to follow the process to complete most individual tax forms and now get adjacent questions answered with little extra effort. Notice that the platforms did not try to be something they are not, but they simply enhanced their mission’s execution with AI as a component of the story.

Number 2 – Strong guard rails. It’s clear that they have constrained the capabilities in this version to what they could control. This is especially true for Intuit, which is pretty weak at answering generic questions, but I found the workflow excellent. The Intuit engine will act as part of the validation engine for your actual return and interrupt to ask required information. This was a great experience and was better than last year’s rules-engine alone process. It isn’t great at retrieving true Q&A, such as “what are the tax rules for paying my kids for participating in a puppy training business”. When building a platform for customer-centric questions a business needs to weigh the balance between answering questions and answering them correctly.

Number 3 – Competition in Software Digital Platforms. As to be expected, the Software Digital Platforms companies are the fastest moving in adopting AI, closely followed by professional services and finance. In this case all three are present and accelerating the time to market. It’s clear that both vendors believed that they needed to bring something to market this year, even it is constrained and offers very targeted value propositions. Both will learn from this and continue to iterate over time. Both know that AI copilots and ultimately agents will need to be part of their strategy. Lack of engaging in the space would make the business seem less capable and easy fodder for the other business to claim market share. Companies in nearly any agency should take note and be the leader in their space.

Number 4 – In-Flow Experience. The experiences didn’t feel like they existed simply as an add-on to the process, but were value-adds and even better, were truly and integrated part of completing the return. Good job at understanding the job-to-be-done and the relationship to the use of an AI platform. In one experience after a missed tax element was not completed, the Intuit engine stopped the flow and prompted for the missing data, while adding helpful additional information.

Number 5 – Connection to Real People. The experiences have the ability to escalate to an engaged human who can help with the return. This is a great example of how humans are force multiplied as a result of AI. If the chatbot couldn’t answer the question, the option existed to send the question to a person who can participate right where it left off. Customers don’t like it when they progress through a flow, only to have to start over. The idea of moving forward not backward is key to making it real.

Number 6 – Performance. The platform experience for each was fast and intuitive without a feeling of lag that caused a negative user experience. In building Gen AI solutions we’ve found performance management to be one of the key requirements. The experience needs to integrate the acceptable feeling of delay into questions and answers, similar to chatting or texting, vs. it being completely in-line to the typing experience.

So, what’s next in the platforms?

1. They will increasingly take on agent actions, not just Q&A. As platforms continue to advance they will have the ability to take input and perform an action vs. just provide information back.
2. They will become less constrained as anti-hallucination (or ungroundedness) capabilities become stronger
3. The process will be better at predicting opportunities for tax savings and identifying areas to look. There are many parts of the tax process that needlessly complicated and follow a basic if-then-else rules engine. The AI engines don’t need to do it this statically and can lead down more flexible and beneficial patterns.
4. Enhancement of increasingly capable business tax platforms. The initial experience is more limited to personal taxes in normal situations. The emerging capabilities will begin to disrupt simple business taxes that are becoming a key part of the creator economy.
5. Direct integration into financial tracking platforms that provide key differentiators. The best tax platforms can already input transactions and your income, but the better apps will be able to predict the right choices inside the current year, making it more of a multi-month experience instead of just a once-a-year transaction.

What is my take-home point from all of this? Now is the time to think about the mission of your business and how AI can help your strategy accelerate. Do it responsibly and do it well.