PredictAP Blog

Why accounts payable is the ideal use case for AI

Written by David Stifter | Jul 11, 2024 5:30:28 PM

Artificial intelligence (AI) is top of mind for real estate companies. The internal conversations have shifted from “if” they will incorporate AI into their operations towards “when” and “how.” Amid rising pressure to make sweeping changes and eke out new competitive advantages, more real estate CTOs and COOs are getting a clear mandate for AI adoption.

Of course merely checking the AI box isn’t enough—you need results. With a new tool it can be tempting to try it on everything, but getting a clear win to establish credibility and cement buy-in from the business. Accounts payable (AP) is the ideal place to start to secure your first AI win. 

The biggest obstacles to any digital transformation efforts are usually organizational resistance to change and lack of clarity around ROI. For AI specifically, lack of in-house expertise and availability of clean, structured, well annotated training data are top concerns. 

Why AP is the ideal use case for AI

There’s significant pain around AP data entry.
While AP automation has largely focused on building and automating approval workflows, entering invoices into the system of record and coding them for payment has remained a frustrating, mostly manual process. The work requires significant company specific institutional knowledge, but is repetitive and time consuming, and hard to hire for. Turnover is incredibly risky, and new hires take a long time and lots of supervision to ramp up.
 
You already have training data

AI learns best from data that is highly structured, well annotated, and plentiful. Your entire invoice and payment history captured in your accounting system of record or ERP is perfectly suited for training a machine learning model. An AI-powered invoice capture solution can identify patterns in your invoice history and use them to automate invoice coding based on predictable scenarios.

There is a review step before payment approval.

While it helps to have historical data to train your AI, for a complete long lasting solution you need an efficient feedback mechanism to help it keep learning. AP typically always has a person reviewing payment and adjusting for changes in the portfolio or accounting treatment. This is an ideal way to continue to learn each day as the business grows and evolves. If done properly, that existing payment review process becomes the source of real time retraining of the AI model without creation of a new task like a template maintenance function or group, or IT needing to re-record RPA instructions. It can dynamically adjust without additional human intervention.

The ROI ties back to real money.

The cost of manual AP is high: there’s labor costs that increase when you need to add headcount; inadequate training and poor visibility can lead to duplicate and fraudulent payments, or incorrect allocations that cause budget issues later on; when property managers also handle vendor invoices, the time they spend on administrative and data entry work is time taken away from higher value work; missed or late payments lead to late fees, frustrated vendors, and impacted credit ratings. Solving AP data entry and coding with AI will have a measurable impact on operational costs and budget integrity.

There’s no need to build it yourself.

Targeted applications for AI exist today with a focus on specific problems. There’s AI for financial modeling, lease abstraction, and yes, accounts payable. You can implement a best-of-breed targeted AI solution without kicking off another multi-year project.

Manual AP work is repetitive and arduous, and creates little value despite being mission critical. Using AI to automate it unlocks the full value of past work and existing data, and centralizes the “tribal knowledge” in a way that makes it actionable for the organization and less dependent on individual employees’ memory.

Why AI is well suited for accounts payable

AI can learn and adapt.

If templates were a good solution for AP work, there might be no need for people on the job, let alone AI. But even if 80-90% of invoices follow a predictable path, there is enough variability that requires human judgment and problem solving, and it’s impossible to anticipate and templatize for every single scenario. Your accounting rules and codes can also change over time—especially as your property portfolios evolve. AI can tackle the 80-90% of predictable, repetitive work to free up time, and continue to learn and evolve as the rules change, without requiring templates, rebuilding, or reprogramming.

Narrow, deep focus delivers a faster time to value.

Despite impressive advancements in generative AI, we’re nowhere near a generalized artificial intelligence that can run your entire company on auto pilot. AI solutions are popping up everywhere, but it can take some trial and error to get a good outcome. AI that is already trained for financial, legal, or other types of data allows you to move quickly and in a more impactful way. Thinking smaller will help prove viability, unlock ROI, and temper resistance to change.

AP is the perfect place to start your AI transformation. You can leverage existing data, capture and safeguard institutional knowledge, create added visibility, and free up your team to focus on the work that only they can do. 

If AI adoption is your next priority, start with a clear and present business pain and a targeted approach to earn your first win, and build a model for success you’ll be able to replicate across the entire organization.