Part of the beauty of generative AI is it can work with what you already have in place, and therefore it can cost substantially less. To make costs even more palatable, many automation and AI projects can qualify for Section 41 R&D credits, which can be used to offset development costs.

For businesses to fully and immediately reap the return on investment of AI, there must be a methodical approach in choosing the right solutions and the right partners to implement them. Any technology investment strategy must take into account potential offsets from tax incentives, allowing a business to get immediate returns on their transformation. Below is a primer on the types of generative AI projects that qualify for R&D, as well as projects that may not qualify but still can provide an immediate return.

R&D credits for generative AI

While Congress has not yet explicitly incentivized businesses to adopt AI and intelligent automations, the R&D tax credit has been updated over the years to reward businesses for deploying advanced technologies. The credit remains the largest tax incentive for American businesses, and those investing in AI have several potential pathways for generating substantial refunds that can be used to offset the costs of implementation.

For an AI solution to qualify under Section 41, it needs to be a new or improved business component (software is generally a qualified business component) that is either sold as a product or used by the taxpayer in a trade or business.

Business intelligence through AI is particularly ripe for generating credits. As a simple example, for most businesses, CPA firms included, data often exists across multiple systems that don’t talk to one another. Usually, it requires a person, or team of people, to manually input data from one system into another to stay current and maintain consistency. If instead a firm were to employ intelligent automation or generative AI to compile and sync data across all systems to better serve clients, that not only would dramatically increase efficiency but also could qualify for the R&D credit. Additionally, if AI were employed to generate reports and insights based on that data, the implementation costs could potentially generate additional credits.

The most common expense that drives the credit is wages of employees working on qualified projects, such as internally employed software developers or data engineers. What is sometimes overlooked is contractor-driven credits. That is, if you hire an outside contractor to develop an AI solution, you can still generate credits based on 65% of that contractor’s costs.

It should be noted that Section 41 would not apply to generative AI use cases that are for “internal use” only. That is, if the AI solution is developed for general and administrative functions, it would not qualify. The above example could qualify because it would be developed to provide services in the firm’s trade or business.

There are of course other scenarios where businesses can get immediate returns on generative AI implementations that may not necessarily translate into credits. Just because an AI solution can generate a credit does not necessarily mean that should be the priority initiative. It’s important for businesses to strategically develop a roadmap for deploying AI.

The substantial transformative potential of generative AI

While return calculations are still in their infancy with this emerging technology, the sheer volume of ways returns can be found on this investment are encouraging. Here are potential business benefits generative AI has demonstrated, the specific use case examples that can lead to immediate value, and which of these can potentially qualify for R&D:

Cost efficiency: Imagine trimming your operational costs by automating complex tasks and processes — generative AI makes it achievable. By taking over manual-intensive activities with generative AI and intelligent automation centric designs, companies can reallocate human talent to more strategic initiatives, thereby optimizing both human resource allocation and operational expenditure.

We worked with a tax firm using generative AI to read its client information and add relevant data to their work papers automatically. The algorithm could read and identify the data and properly classify it into each workpaper. When the algorithm was uncertain, it would verify the information with a tax professional and continue to complete the workpaper when resolved. As a result, 15% of the time was saved compared to manual data entry, allowing for speedier tax return preparations and more time to review the workpapers and look for additional strategic tax opportunities for each client.

While this provides an immediate benefit, this likely falls under the internal use exception and would not qualify for the R&D credit. That said, the value here would be immense, and credit aside, most CPA firms would jump on having such a tool.

Enhanced personalization: Generative AI is setting new standards for customer engagement, offering tailor-made content and experiences that captivate and retain clientele. In the era of personalization, generative AI is an accelerator of customer experience differentiation.

One firm we spoke with in the asset management sector is using generative AI to quickly create customized financial reporting, analysis and recommendations tailored to individual client needs. By analyzing the financial data and preferences of clients, an AI algorithm trained on a firm’s past work can generate personal insights such as budgeting suggestions, investment strategies and tax optimization recommendations.

The customized reporting and analysis is a borderline R&D credit candidate. Simply deploying off-the-shelf AI to do analysis would not qualify. But if the firm had to write code and build automation to compile the underlying data and then build AI to do the reporting and analysis, then it could possibly qualify for the credit.

Acceleration of go-to-market timelines: Time waits for no one, especially not businesses in competitive markets. Generative AI acts as an accelerator, enabling companies to conceptualize and materialize ideas at unprecedented speeds, reducing time-to-market and increasing reactive agility to consumer demands.

One firm we worked with has used generative AI to significantly improve the speed at which it builds go-to-market messaging. The industries it serves are highly regulated and, as regulations change, the message and services change rapidly with it. When these changes occur, new marketing and client messaging could take up to a month to produce and deploy. With generative AI, the response time from concept to message has been reduced to under 72 hours, allowing the firm to be first to market with new concepts within its competitive peer group.

While there is immediate value here, this is also a clear example that likely would not qualify for R&D as it would fall under the internal use exception.

Quality uplift: Quality can now be quantifiable, predictable and improvable, thanks to generative AI’s capabilities in pattern identification, predictive analysis and automated testing that streamlines quality assurance frameworks to new heights.

A firm we spoke with performed complex tax calculations that included significant amounts of data and multiple workpapers as part of the final report. Manually reviewing these workpapers was time consuming and error prone. Using generative AI, the firm was able to create an AI review bot that recalculated each workpaper, tied it back to the source information and made sure the workpapers information tied out with the final report. This additional quality review significantly lowered the error rate and gave the manual reviewers more time to work on areas of the review that required significant judgment. Again, this is likely internal use as it is administrative in nature, but the improvement in efficiency and quality cannot be understated.

Streamlined customer experience: Generative AI can also be used to enhance the customer experience. One of the best use cases we’ve helped firms with is automating requests to clients for documents. Automations can not only send requests for documents but then also read the documents, process and catalog them, and then notify the client if additional information is required.

Critically, software developed that allow third parties to initiate functions or review data on a taxpayer’s system are not considered internal use and therefore potentially qualify for the R&D credit.

Minimizing investment risks with the R&D tax credit

The breadth and potential of generative AI is too great for the accounting industry to ignore. Early adopters of this technology are going to find themselves with topline growth, more efficient delivery and higher quality service and deliverables to their clients.

The R&D tax credit can soften the financial impact of integrating generative AI, providing peace of mind and the fiscal courage to invest in what could very well be the defining technology of the next decade. The confluence of generative AI’s boundless potential and the strategic application of the R&D tax credit can be a game-changer for businesses willing to lean into innovation. Through judicious planning, thorough documentation and professional guidance, companies can bolster their generative AI initiatives, positioning themselves as pacesetters in a tech-driven economy, all while having their financial back covered by the R&D tax credit.