From Macros to LLMs: The Evolution of Sales Tax Automation
Automation has come a long way since the days of Excel (or Lotus 1-2-3) macros. Depending on how long you’ve worked in tax or finance, you may recall a time when automation meant building spreadsheets, running basic scripts, or manually updating data across platforms. This isn’t to say that these manual processes have been completely replaced, but they are evolving.
Today’s AI tools might feel like a sudden leap. However, they are actually the latest step in a long progression that has steadily reshaped how we manage complex tasks.
Still, no matter how advanced the tools become, one element remains constant: human oversight. Automation may be designed to reduce the burden of repetitive or logic-heavy tasks, but it does not eliminate the need for human judgment. In fact, the concept of a “human in the loop” underscores the crucial role people play in guiding and enhancing AI systems.
If you’re curious to discover how AI can fit into your sales tax practice today, make sure to join our upcoming webinar on June 26th, Harnessing AI for Sales Tax: Reduce Risk & Improve Efficiency for Your Business.
Before we explore where AI is headed and how it is transforming sales tax, let’s take a moment to reflect on where automation began and how far it has come.
A Timeline of Automation
To fully understand where we stand today with AI in sales tax, it is helpful to look back at how automation has evolved over time. What we now consider cutting-edge began with far more labor-intensive tools that laid the foundation for today’s innovations. Each wave of automation represents a step toward saving time and improving accuracy. Understanding this progression can help businesses better evaluate where AI fits into their current workflow—and where it’s heading next.
Basic Automation in Sales Tax (1980s–2000s)
In the early days of automation, professionals relied heavily on spreadsheet software to manage compliance tasks. One of the pioneering tools of the 1980s was Lotus 1-2-3, a dominant spreadsheet program that allowed users to combine data, formulas, and rudimentary graphics. For tax professionals, it provided a way to organize rate tables, build manual audit logs, and manage exemption certificates long before specialized tax software existed.
By the 1990s, Microsoft Excel became the industry standard, and its macro capabilities allowed teams to automate simple yet time-consuming processes. Users could create formula-driven templates to calculate tax rates by jurisdiction, summarize transactional data, and flag discrepancies. For example, one common approach was to use a VLOOKUP function to manually match customer locations to tax rates entered into a separate sheet.
Repetitive tasks, such as copying and pasting large amounts of data, formatting tables, or generating summary reports, became faster through the use of basic scripting and keyboard shortcuts. However, this kind of automation still demanded a high level of manual oversight.
Tax engines were limited to Vertex and Taxware (now known as Sovos). Functionality was basic, jurisdictions were limited to the US and Canada. And a different version of the tax engine was required for each different type of selling or financial system. Everything was on premise which incurred significant technology costs.
The biggest drawbacks of this era were its limitations. Spreadsheets were prone to human error, lacked consistent version control, and offered little in the way of auditability. Additionally, they highlighted the need for more scalable solutions as the demand for communication between complex platforms increased alongside technological advancements.
Sales Tax Add-Ons and Middleware Tools (2000s–2015)
The onset of the 2000s brought increased internet adoption and, consequently, changing consumer expectations. Alongside these changes, businesses began transitioning to more complex digital environments, utilizing digital commerce and platforms. This created a more complicated landscape for sales tax but also introduced the challenge of data silos and software sprawl. The need for seamless data transfer between systems became increasingly crucial.
Professionals wanted their systems to work together, which spurred the creation of software integrations and middleware tools to allow data to flow more seamlessly from one platform to another. Frameworks like ASP.NET, developed by Microsoft and released in 2002, are a prime example. ASP.NET allowed developers to create custom applications that could connect directly with ERP systems, e-commerce platforms, and tax engines.
Another important tool during this time was Zapier, which first debuted in 2011. This platform allowed non-developers to connect different web applications through pre-built workflows. Although not explicitly designed for tax, Zapier has provided tax teams with more flexibility in automating parts of their compliance processes, eliminating the need for heavy IT involvement.
Sabrix (now known as OneSource Indirect Tax) hit the market and broadened the thinking of what sales tax functionality should include as well as moved the industry forward technologically. Other more simplistic tax engines, such as those offered by providers like Avalara and TaxJar entered the market as e-commerce grew. From automating tax rate lookups to jurisdiction assignment and calculations at the point of transaction, these engines were increasingly integrated into ERP and shopping cart platforms to make tax compliance faster, more accurate, and less reliant on manual intervention.
Despite these benefits, the fact that these automations were rules-based meant they needed constant updating to keep up with changes in tax laws and rates. Without oversight from tax professionals, discrepancies could be missed, meaning that automation still had a long way to go before it could be trusted to operate independently.
Bots and Scripted Processes (2015–2022)
Building on the logic of macros and earlier rule-based systems, bots introduced through Robotic Process Automation (RPA) brought a new level of efficiency to manual, repetitive tasks. Unlike Macros that were limited to working within a single application, these bots could be programmed to perform highly specific actions across applications, such as logging into state tax portals, navigating to a particular file, downloading it, and then entering the data into another system.
A major benefit of bots was the way they reduced the time spent on tasks that previously required significant effort from humans. These automations improved turnaround speed for routine filings and efficiently pulled invoices and other documentation needed for audits.
While these advancements were significant in comparison to the early days of automation, they were still far from perfect. The rigid, preprogrammed paths were a clear limitation, as they meant the system was unable to adapt to change. Bots would often break if a state website updated its layout or login process. Even minor deviations in a process could cause a bot to fail at its task, which consequently required manual reprogramming to create a new path to success.
While bots brought clear gains in efficiency, they also exposed the need for more adaptive tools. The bots set the stage for the introduction of artificial intelligence, which offered more than just automation but also learning and responsiveness to change.
Practical Uses of AI in Sales Tax Today
Following the rise of bots and scripted tools, the introduction of large language models (LLMs) such as OpenAI’s GPT-4, Google’s Gemini, Microsoft’s CoPilot and Anthropic’s Claude marked a major turning point for automation.
In the world of sales tax, these tools began reshaping how tax departments handle research, product classification, document review, and user support. Most of these LLMs debuted in late 2022 to early 2023, with subsequent updates released regularly even to today!
By the midpoint of 2024, leading tax vendors began embedding generative AI into their platforms. Using these tools in your daily life still requires effort, mainly in the form of proper prompting, to train your chosen platform to understand your needs.
Some practical use cases for AI include scanning invoices, emails, and contracts for taxability clues. For example, AI can identify taxable digital services from vague invoice descriptions. Predictive models can flag high-risk transactions or anticipate audit triggers. LLMs trained on jurisdiction-specific tax rules offer near-instant guidance on complex questions.
Advanced tools can even search SharePoint repositories, extract invoice details, and automatically create structured reports. The ability to build custom GPTs enables you to increase security, but they also allow you to tailor your model to a specific workflow.
AI can interpret vague language, classify data, and generate insights, but it can also produce incorrect results if inputs are poorly structured, misunderstood, rely on inaccurate information or hallucinate because of lack of available information. It’s essential to remember that, despite its potential to improve day-to-day tasks, AI is not a silver bullet. With careful supervision, governance, and human judgment in the loop, tax professionals can use AI with greater confidence and control.
Unlock AI’s Potential for Your Business
From the earliest days of macros and spreadsheets to the middleware boom of the 2000s and the rise of bots in the 2010s, each phase of automation has steadily built toward today’s AI-powered landscape. The tools we now consider advanced—like GPT-4, Gemini, Co-Pilot and Claude—are just the beginning.
The pace of advancement is accelerating. In 2023 generative AI was already progressing twice as fast as experts predicted, according to a study conducted by McKinsey. The current trajectory for AI aims to not only assist with tasks but anticipate and adapt to user behavior. With new capabilities emerging every few months thanks to LLMs consistently learning and leapfrogging each other, we are likely to see even more tailored and integrated AI solutions within the next year.