T-Bank — SBA Paycheck Protection Program loan submission, automated end-to-end

The challenge

T-Bank had to maximise the number of PPP applications submitted before the federal funding pool ran out. The SBA submission portal was a textbook overload scenario: hundreds of community banks submitting at the same time, and the site went down or threw errors for minutes at a stretch. Every crash forced operators to re-enter the same data — bank details, business EIN, owners, payroll numbers — from scratch.

A simple per-application loop took 10–15 minutes when things worked. With error rates and retries factored in, a single operator could only submit a handful per hour. The bank had hundreds of applications waiting.

The solution

We built an attended UiPath robot that ran alongside the operator. Application data lived in a CSV file generated from the bank's loan platform; the robot read each row, navigated the SBA portal, and populated the form in seconds.

Crucially, the robot was crash-aware. When the SBA site froze or returned an error, it paused, waited for availability, then resumed from the exact field it was on — no re-entry, no duplicate submissions. Operators only stepped in for true exceptions (missing data, payer-side rejections, etc.).

Total build effort: roughly eight hours from kickoff to a working robot in the operator's hands. No internal IT involvement was required — we configured it on a single workstation and let the team scale by simply running more instances.

The outcome

Submission time dropped from 10–15 minutes to 6 seconds. Across the program, the bank submitted over 1,000 applications totalling ~$100M, generating roughly $5M in fee revenue for T-Bank.

More importantly, the bank captured PPP volume that would otherwise have been lost to the funding cap. The case became Flobotics' reference engagement for time-critical, attended-RPA work in regulated finance.