AI, automation, and the future of accounting
AI, automation, and the future of accounting
Kurtis Hanni is a CFO, consultant, and podcast host who writes about business finance for his newsletter “Frameworks & Finance,” which helps more than 19,000 people understand their financial statements and make strategic decisions that drive business forward.
In 2013, a study from Oxford University came out that concluded there was a 98% chance the jobs of bookkeepers and accountants will be automated.
At the time, this possibility seemed like a decades-off dream.
But as AI has advanced over the past six months, winds seem to be shifting.
I’ve used a lot of different solutions, taken a lot of different demos, and been left with the feeling that while these technologies are helpful, they’re not replacing any jobs.
Let me explain using QuickBooks as an example.
Anytime I buy the next item at Amazon, it wants to code it how it was previously coded. It’s using a simple rules engine that says, “code it off the last transaction.” While this is helpful some of the time, it’s frustrating more often than not.
Accounts payable platforms use OCR to pull vendors, amounts, and dates from invoices. When you combine with vendor-based rules engines, these platforms can get the coding right much of the time.
But, it’s still just wrong enough that someone has to interact with every transaction. With APIs and integrations between platforms, companies can now pick the ideal platform for what they’re trying to accomplish and use these workflows to get the best of all words.
They might choose QuickBooks for accounting, Fathom for reporting, Brex for employee expense management, and Gusto/Rippling for payroll and HR. While most of these make your life significantly easier, you’re still having to extract the valuable parts of each of these sources, leaving you with a mishmash of solutions that don’t talk well together.
But with the introduction of AI, this is slowly changing.
Brex has partnered with OpenAI to create chatbots that make interacting with the data, and extracting insights is easier than previously imaginable.
While each technology is novel and interesting on its own, it's when these technologies merge that things become interesting.
As AI becomes more powerful, these previously clunky interfaces suddenly become smooth and intuitive. As technologies become easier to patch together, integrations become seamless. It’s in these synergies that these technologies start adding real value.
I’m building my fractional CFO firm around these principles: I want to help businesses automate what they can so that they can get quicker access to data, which allows them to get quicker insights, which will result in better business decisions.
It’s in these applications that we start to see the benefits of AI and automation. Seeing the advancement of AI over the last six months, I can finally see more clearly what this future will look like.
The impact of AI & automation
Even the best data entry clerks need breaks. AI and automation on the other hand can work around the clock. They also don’t make mistakes or get interrupted.
Just after graduating college, I found myself entering large amounts of data manually because the team I was on was way down IT's list of priorities.
We’d spend hours each day, sometimes working overtime, to get the transactions in the system in a timely fashion. Months later IT finally automated an import that significantly reduced the amount of work we were having to do. With improved Automation and AI, those months turn into immediacy.
Natural language processing and APIs can immediately turn unstructured data into structured data that can be moved from system to system, getting rid of previously manual processes.
One point of pride for me and a buddy is that we alone were quicker and more accurate when entering transactions than a whole team of 10. Even with this, our error rate was dramatically higher than a well-trained AI’s will be.
Today our data is limited by our ability to enter data accurately and track down data we’ve messed up. Auditors have to choose samples (or small sets of data) to review because the whole dataset is typically too big.
With AI, you will eventually be able to automate the analysis of whole datasets and, at the same time, improve your ability to identify unusual transactions. This will not only transform the audit process, but also mean that businesses can more quickly detect fraud, duplicate payments, or entry errors.
The result? Better books save time and money.
When transactions can be processed and coded instantly by your software, financial statements and accounting reporting can be delivered immediately, too.
I promise accountants aren’t intentionally slow. But when preparing financials, we want to make sure they’re right. We review transaction data, reconcile accounts, and then do our analysis before passing them on. But now, much of that will be able to be automated, meaning financials are produced in the snap of a finger.
This drastically changes the ability to forecast, as the speed makes them more fluid and easier to interact with. Instead of having to create a fragile Excel model, you’ll be interacting with an intuitive and friendly software interface.
Quicker access to more accurate data means decisions are being made on what’s currently true versus what was true last month. In theory, this should improve your decision quality, as you can act more confidently and quickly on trends as they emerge.
There will be a counter side to this: you need to avoid overreacting to small bits of information, which could be hard for certain already volatile people.
I spoke to the team at Brex, which is deploying cutting-edge large language models (LLM) to supplement their platform, and here’s Brex Co-CEO Henrique Dubugras’ take:
“The goal is to make more financial data and decision-making happen in real time. But the ‘data everywhere’ model isn’t helpful. It defeats the purpose if you have to layer AI into multiple financial tools and still piece it all together manually. A unified spend platform allows you to better trust the analysis from a next-gen AI solution because it’s a more holistic accounting of your spend.”
With AI-enabled finances, your finance leads become a real-time partner in decisions, with the latest and greatest information. I cannot wait for this day!
AI and automation mean all companies, and individual company shareholders, can get their preferred reporting and KPIs served up to them as they want to see it.
Using data captured from other industry partners, AI can serve up suggestions that will closely align with what each might want to see. Then, by giving it your preferences, you not only get the data you want, but you get it more quickly.
No longer will waring factions of management have to fight over the ultimate reporting packet. From here on out, everyone gets what they want when they want it.
Future AI will become a copilot to your finance team. It will be able to complete tasks and analyze data at a rate no human can duplicate.
When I go look at historical data today, I only have high-level numbers. I have to dig into different reports and data to dig into and get deeper insights.
With AI trained on your data, retrieving that data is as simple as asking “Pull revenue by service type for the last 10 years… Now reclassify Vendor A as X service type and rerun the report.”
“Copilot AI” could also make suggestions based on industry standards it finds on the web. This level of personalization and customization will completely change the way you work.
Will AI/automation replace my accountants?
For now, I say no. I think the initial impact we see will be retraining. These tools are not “there” yet and will still require significant human intervention. They’ll also require time for businesses to adapt to how they operate and get comfortable trusting their output.
Along the way, we’ll see some bad solutions and bad outputs, which will lower trust. But over time we’ll learn from these issues and see them slowly start replacing previously manual processes.
So, if you’re an accountant, what does this mean? Here is what I think the future of accounting looks like:
1. Accountants will consult. AI can’t make decisions for you. and it will miss the additional human context. Accountants who build an internal or external “consulting practice” will be in high demand.
2. Accountants will specialize. In a world of powerful AI tools, generalists become less valuable. AI tools can and will replace generalized knowledge, but specialized knowledge (where consultation is needed) will always be in demand. I’m not going to use AI alone to analyze an acquisition target. Great M&A attorneys and accountants bring way more to the table than AI ever could.
3. Accountants will become AI-aided. Quit looking at AI as a threat. If you embrace the tools, you can instead focus on higher-value tasks. Those who know how to work the software/AI are even more valuable than they were before.
4. Accountants will still be a face and listening ear. No matter how much we love AI and automation, humans trust humans the most. Maybe there will be fewer jobs, but if you can master the interpretation and communication of the numbers, you will always have a job.
This article originally appeared on Frameworks & Finance.