How Neo.Tax Solved the R&D Qualification Process

As anyone who has filed for an R&D Credit knows, the key challenge is to determine which of your company’s projects meet the qualification criteria of the IRS's 4-Part Test. To qualify, each project must:

1. be technical in nature

2. be created for a qualified purpose

3. work to solve a problem with technical uncertainty

4. involve a process of experimentation

In order to answer these questions for each and every project your engineers have worked on over the course of a tax year, tax and accounting teams often interview the engineering managers and ask them to try to remember details about the work. As you’d expect, relying on human memory leaves much to be desired for the IRS; more than that, the entire process is a huge time commitment for both the tax and the engineering teams.

Neo.Tax’s Automated Approach is Different

Neo.Tax’s approach is much more rigorous and thorough than traditional ways. Better yet, it works at scale across hundreds or thousands of projects.

Here’s how our AI-powered system works:

1. Our LLM reads through every single ticket and task within your company’s project management software.

2. It evaluates each project’s eligibility step by step in the 4-part test using a tailored model, trained on IRS codes, regulations, court cases, memorandums, letters, and guidance.

3. Finally, it creates a detailed narrative of each project, explaining why or why not it qualifies for the R&D credit. 

4. You can then inspect each step, and determine whether you agree or disagree with Neo.Tax’s determination.

5. By delivering the detailed reasoning for each determination, Neo.Tax creates an audit-ready study automatically.  

Rather than relying on interviews, often conducted months after the work was completed, Neo.Tax’s approach uses contemporaneous documentation to create a much more accurate filing.

‍

How Neo.Tax Solved the R&D Qualification Process

Qualification is broken down step by step and our automated system evaluates each aspect of the 4-Part Test in order before proceeding.

Qualification is broken down step by step and evaluates each part of the 4-Part Test separately before proceeding to the next step

Because the model has been trained on millions of tickets across dozens of R&D industries, it understands a wide range of technical jargon, which it uses to determine whether a project is Technical in Nature. 

For example, it can digest sentences like, “FailPoint objects are designed to be defined as nonlocal objects” and “Review ad copy on AXC-15 iteration” and automatically determine that the former is Technical and the latter is Non-Technical. 

Next, it moves onto the Qualified Purpose section, and evaluates every project summary and ticket detail against the IRS code and regulations. When a project involves Internal Use Software, it automatically applies the additional high threshold for innovation test.

Best of all, rather than make a 50/50 call, the automated system flags uncertain cases for follow-up and human review.

As it moves onto the Technical Uncertainty question, it begins by examining each project to identify major unknowns at the outset. Then, the LLM searches through individual tasks/tickets to identify supporting evidence. 

Finally, Neo.Tax examines project details to identify each project’s approach to a problem to determine if there was a Process of Experimentation. That means searching through individual tasks/tickets to identify supporting evidence of hypotheses, discarded paths, and iteration towards discovery.

‍

Why AI Actually Works for R&D Tax Credits

In this moment, when AI is being put in everything, it’s easy to be wary of an AI-Powered Fix. But the problem that we set out to solve was a pressing one. The old way of filing R&D credits is an inexact (and expensive) science. The process takes months of work for tax teams, and relies heavily on interviews with engineers months after work was completed. It’s an arena where guestimates ruled the day, but the IRS has been signaling that the burden of proof will be rising rapidly for R&D filers. 

Neo.Tax’s AI-powered system uses data that exists within your system, but that no tax team (or even accounting firm) could ever sort and digest in time for Tax Day.

Large bodies of data that need to be sorted based on extremely specific and voluminous rules is what LLMs were built for. And that’s what we built Neo.Tax to do. It digests your project management software and weighs against the entirety of the relevant tax code and IRS guidelines. It’s more accurate than fuzzy memory and it’s scalable across thousands of projects.

And this point bears repeating: there’s no blind faith involved. Every determination is easy to inspect and adjust within your management context. And the extensive project narratives, built off your contemporaneous data, mean that the study Neo.Tax delivers is audit-ready.

This is the new faster, cheaper, more accurate way of filing for R&D credits. So, find out what you’re owed today.