Why and how to use AI to Claim SR&ED tax Credits

Ever since artificial intelligence (AI) tools such as ChatGPT and Claude powered by LLMs have become widely used, entrepreneurs have expressed concerns regarding the use of AI to claim SR&ED tax credits.

The tax authorities, namely the Canada Revenue Agency (CRA) and Revenu Québec (RQ), analyze whether taxpayers qualify for a tax credit such as the SR&ED by answering two questions.

1. Do the facts provided meet the criteria set out in the tax law?
2. Has tangible and contemporary evidence been submitted to corroborate these facts?

Ultimately, the Tax Court of Canada also asks these questions, where required.

The answer to the first question is essentially a line of reasoning. Typically, when claiming SR&ED tax credits, you must demonstrate that, in fact, you attempted to resolve a technological uncertainty through a systematic investigation. 

The second answer is based on the organized production of documents that corroborate the facts presented in the argument (timesheets, test reports, data collection and financial documents, for example).

Every claimant knows that considerable work must be undertaken to meet these two requirements.

AI can help you gather evidence (question 2)

In 2026, this work is no longer carried out as it was as in 2016. AI can be a valuable asset when responding to the second question by organizing and structuring the evidence (which will support your reasoning) based on your organization’s contemporary documentation. In this context, AI can help you by leveraging its aggregation, automation, classification and even statistical and mathematical capabilities.

A few well-formulated prompts could save you from spending endless time on tedious tasks. For example, you could write a prompt for Claude or another AI tool based on the example below.
 
 “Generate an Excel document that summarizes the results compiled in directory ABC between January 1 and December 31 that focuses on the maximum and minimum values. Analyze the timesheets of resources who participated in producing and analyzing this data.” 

The results generated by such queries will allow you to draft a detailed and structured summary that reflects your organization’s activities. 

Only humans can deliver a clear, coherent and credible line of reasoning (question 1)

The documents that you submit as evidence must be supported by a line of reasoning that the tax authorities will find coherent and credible. This element is fundamental. Striving to use AI to prepare an argument is a strategy (or ploy) to have AI analyze the data with a view to ensuring a favourable tax ruling… regardless of the information that is shared with the tool.

Therefore, the line between facts and a debatable interpretation of those facts inevitably becomes blurred. The tax authorities, therefore, have reasonable grounds to doubt the absolute credibility of a claim, which most often leads to a de facto denial without the possibility of further discussion.

While AI has the capacity to classify data and infer from that information, it has not yet achieved the level required for interpreting legislation and taxation rules. While certain individuals believe this ability is within reach, they are mistaken. While using AI during the initial stage of filing your SR&ED claim to aggregate and organize the supporting evidence is wise, useful and efficient, AI cannot yet replace human intelligence when it comes to crafting a comprehensive and sound argument that will be considered admissible by government authorities.

Author’s note: No AI tools were used to draft this notice!