Point of View

SBA and RPA: Could the real badly-behaved bot creators please stand up? Everyone is saying “It wasn’t me”

May 14, 2020

As we’ve said of late, the digital workforce that wasn’t finally has its burning platform and a chance to shine. Since global lockdown commenced, we’ve been tracking, talking, and learning how enterprises and their service and technology partners are using automation to help them function. In recent weeks, one of the most prevalent use cases is in the banking and financial services sector, where lenders are using automation, largely RPA, to support loan processing for government-backed loans. RPA is helping lenders grapple with massive loan volumes and submit them quickly so that they can approve and disperse loans.


The SBA prohibited lenders from using RPA. And there is one big lesson for practitioners to learn from this, namely that automation cannot live in a vacuum.


Insert monkey wrench


Demand overwhelmed the US government’s Small Business Administration (SBA) application and approval portal when the government made a second tranche of funds available on April 27. The purpose of the funds is to support loans for small and medium-sized businesses (administered via banks) as part of the Payroll Protection Program (PPP) during the COVID-19 pandemic. The SBA E-Tran system receives the loan applications and then the SBA processes and approves them. The system struggled to handle the new surge of applications and repeatedly crashed. As of Tuesday, April 28, the SBA prohibited the use of RPA bots in the application process, but the API route remains open.


Generating much controversy last year, HFS declared RPA dead. We recommend more holistic approaches with RPA simply being one of the tools in the toolbox—not the only tool. This SBA situation exemplifies HFS’s thinking.


Many vendors were already using APIs, or they quickly adjusted to a hybrid approach


UiPath, Automation Anywhere, Kofax, Pega, and Appian all made announcements in recent weeks, saying they were supporting banks in this application process with SBA’s E-Tran loan processing web portal. We checked in with some RPA ecosystem vendors to see their responses to SBA’s stipulation. While there’s a bit of conjecture about how great each respective solution is and some careful wording, they told us, in summary: “It wasn’t me”, “I would never”, or “I’ve changed”.


  • Automation Anywhere (AAI) built bots to help small lenders process SBA PPP applications on behalf of their small business customers. Unlike larger lenders, most small lenders do not have automated system-to-system integration with the SBA’s XML API. As a result, smaller lenders have been reliant on either manually entering the information or leveraging RPA bots to help them process applications for the significant number of small businesses they support. This is one of the reasons that small lenders were disadvantaged in the first iteration of the SBA’s PPP program. AAI adds that its automation is not reliant on using the UI. Many bots use XML APIs, and AAI had already been working on implementing the SBA PPP E-Tran automation using the SBA’s XML API.
  • Kofax tells us it publishes each Kofax RPA robot as a “Synthetic API” and that this should resolve the issue of overburdening legacy SBA systems with too many UI inputs. Kofax is of the view that RPA executing as an API is much more efficient when operating data migration at scale and among a mixed modern and legacy IT estate.
  • Blue Prism is supporting banks (including many mid-sized lenders) that are using Digital Worker automation as the equalizer to rapidly support SMB clients. SBA’s mandate prohibits the use of RPA GUI interfaces into E-Tran. Blue Prism and its partner Lateetud developed a ready-to-implement solution that uses XML to access E-Tran, a method that adheres to the latest SBA policy. The solution includes digitizing the incoming application, validating it, and uploading it to the SBA. In addition, Blue Prism has extended the capability to other emerging applications, including loan forgiveness, good standing checks, void checks for fraud, OFAC checks, and mortgage forbearance. Blue Prism can provide spy modes for GUI and API integration into systems.
  • Pega approached the problem with a low-code lending application that banks could download immediately and use to enable their customers to apply for the loans in their preferred channel. Pega configured the data in its case management system to support the myriad of APIs from both process stakeholders and banks. Pega offers built-in RPA as an option, but where APIs exist, it is not necessary to use RPA. Pega also worked with T-Systems to develop a system for the Bavarian government to provide emergency COVID-19 financial aid.
  • Appian’s approach is one of full-stack automation capabilities. It was able to quickly change tack and go the API route instead, using standard workflow and API integrations and minimizing impact to customers.


Aside from our direct inquiries to RPA software vendors about the SBA situation, we also know that:


  • Google publicly announced its PPP Lending AI solution, too; it’s another API-based approach in conjunction with a loan-processing portal and loan analytics.
  • Infosys used the break between allocations of funding to work on an API-based approach. Changes to the SBA site made it difficult for RPA solutions to adapt quickly, and lots of small banks were left behind.


So, none of those kids were in the conservatory with the lead pipe when the bludgeoning happened. The mystery of whose bad bots hammered SBA’s system and triggered the pushback remains unsolved.


Coincidentally, another vendor we spoke with recently is addressing unintended consequences of RPA bots causing asymmetric increases in mainframe transaction volumes.


Hostbridge is looking to improve the performance of RPA with a dedicated focus on its performance accessing mainframes. Observing that an increasing request volume can push mainframes to the brink of their processing abilities, Hostbridge proposes its HB.js as an intermediary between RPA and the mainframe to reduce response times and help RPA “play well with others”—perceptive and timely.


SBA’s system faced surging demand for the second round of funding


There was already a queue forming full of those who were unsuccessful in the previous round. API calls, human-assisted processing, batch uploads, and RPA bots collectively formed an exceptional demand surge. The SBA understandably wanted to ensure the loan application process is equitable, especially for small businesses. Banks, small businesses, and the public roundly criticized the fairness of the loan allocation process in the initial round of funding. It is clear from Exhibit 1 that some lenders had higher levels of success in the initial round of applications. One unnamed lender stands out (showing as #15), and we are told it was using RPA.



Exhibit 1: Lender #15 secured a large volume of low-value loans




Source: SBA PPP Loan Report deck.pdf



We’ve corroborated the correlation between RPA usage and smaller lenders and smaller businesses. So, it appears that a ban on RPA was actually more likely to impact the smaller lenders and smaller businesses. But, after banning RPA, the SBA quickly facilitated a designated processing window for lenders with less than $1 billion in assets.


The Bottom Line: This SBA debacle is why you need nuance and a toolbox approach to automation to get desired results.


SBA faces a difficult task processing these applications at the pace and volume lenders submit them, and it is trying to make it fair for everyone that is seeking help with their payroll obligations. There are throughput and technical limitations. It should go without saying that if there is an API option, it should be the first port of call. As noted in a recent blog with a questionable title, RPA was a blunt instrument here—for its speed to solution and ability to swiftly process loan applications—and it worked remarkably well. Too well. It swiftly overwhelmed SBA’s E-Tran system.


APIs versus RPA is not the point. Automation always lives in an ecosystem with upstream and downstream impacts, and these were not adequately addressed by those setting up RPA. A better strategy would have employed a toolbox approach that leveraged RPA for loan preparation and access into legacy systems in the lenders’ shops, APIs for submission, and humans for oversight, with some substantial volume throttling to give the SBA’s system half a chance at doing its job. If attempting to process at scale through the UI, a little consideration for the system on the receiving end of the requests is not too much to ask, is it?

Sign in to view or download this research.


Lost your password?


Insight. Inspiration. Impact.

Register now for immediate access of HFS' research, data and forward looking trends.

Get Started