Robots To The Rescue

Robots are helping Patelco Credit Union ($8.9B, Dublin, CA) achieve new levels of efficiency in its busy mortgage processing department.

That’s the term Vince Salinas and his team use to describe the autonomous programs that are now handling tasks like collecting data, filling out forms, and assigning third-party services such as appraisals.

“We describe them as robots within the system,” says the vice president of home loans for the East Bay credit union.

The credit union went live with the robots on July 1 after completing the final phase of user acceptance testing (UAT) of a system that uses bots from Digital Align in conjunction with its Mortgage Cadence LOS.

Salinas says Patelco now averages 3,563 successful bot transactions monthly, saving approximately 128 hours of human labor at a cooperative that recorded 8.33% year-over-year growth in its real estate portfolio, which topped $3.7 billion as of the third quarter of 2021.

“As the processing turn time decreases, the opportunity to fund more loans increases,” Salinas says.

Here, the home loans VP describes how bots help boost business and how bot champions have helped people and machines learn to get along.

We began by having our people click buttons to initiate a bot running a job so we could determine their effectiveness … now 13 of the 31 bot jobs available are automated.

What are bot champions? Why did Patelco create that role?

Vince Salinas: Team member adoption and trust in the new technology can be a challenge. For example, some might think, If the bots are successful, will I still have a job? or What if the bot fails? or I prefer to do it the old way.

To proactively address these challenges, we introduced bot champions to help mitigate team member concerns. The bot champions participated in all user acceptance testing and, as a result, have become trusted bot subject matter experts among their peers.

What do the bots do?

VS: Repetitive and routine tasks, which includes ordering third-party services such as title, appraisal, flood, and compliance reports, retrieving them, and prefilling information into PDFs on the LOS path.

We also use them to assign loans to defined process flows and to collect data, perform quality checks, update conditions, and add findings to notes.

Describe the transition from manually initiating jobs to automating the entire process.

VS: We began by having our people click buttons to initiate a bot running a job so we could determine their effectiveness. We’ve been increasing automation since then, and now 13 of the 31 bot jobs available are automated.

How has your use of bots changed and progressed since you began using them?

VS: We continue to examine bot performance for timing. We have six bots. If they are all busy running morning jobs, the on-demand [non-automated] bot requests are delayed longer than our expected service level agreement. As a solution, we changed bot timing to include both AM and PM runs. Running the bot automation twice a day shortens the time to complete the job.

Do you foresee bots replacing people?

VS: The goal is to increase loan growth without increasing staff, not to replace people. Bots are good at following commands. Successful transactions average 96% with only a 4% fail rate. But they can never replace the human responsible for decision-making, relationship-building, or learning new skills.

Credit unions are all about member service. How has this use of bot technology helped there?

VS: Member surveys suggest reduced response time has a direct correlation to member satisfaction.

What advice do you have for credit unions newly working with bots or considering adopting them?

VS: First, remember you know your processes, you know the dynamics of your team, and you know the desired result better than a consultant. Avoid giving your consultant complete autonomy.

Be active and present for all decisions. Having an internal subject matter expert dedicated to discovery, process design, and user acceptance testing will reduce unnecessary back and forth caused by misinterpretations of desired outcomes.

Finally, there are network migration updates that can routinely disrupt bot performance. Get IT involved early in the design process to ensure schedules do not conflict.

What’s the No. 1 thing to do or avoid? Why?

VS: When managing a project, we’re inclined to start with the immediate wins, the quick fixes. However, quick fixes can make your process disjointed, without a flow.

To avoid that, avoid the short cuts. Map your journey first and highlight what you want the bots to solve or achieve. Then follow the flow when implementing them. Critical elements for success include a clear vision, a clear timeline, recognizing process gaps early in the project, and early team member buy-in and adoption.

This interview has been edited and condensed.