Paul Margarites knows a thing or two about digital transformation.

As head of U.S. commercial digital platforms at TD Bank, a subsidiary of $1.73 trillion TD Group, Margarites is using his past experience — including mergers and acquisitions at $3.24 trillion former Bank of America Merrill Lynch and wholesale payments innovation at $4 trillion JPMorgan Chase — to help TD’s commercial clients self-automate through a vertical-wide API and embedded finance push.

“We will focus on externalizing our APIs in a safe and secure way to enable our clients to own their banking experience and give them the tools to automate their own processes,” Margarites told Bank Automation News when asked about the bank’s 2022 technology plans.

Margarites sat down with BAN to talk client automation across the bank, along with best practices for deploying novel technologies and automations. What follows is an edited version of the conversation:

Bank Automation News: What are some notable automation projects you’ve pursued in the last 6 months?

Paul Margarites: Notable projects on the client experience front include our work around embedded banking which helps clients automate payables processes as well as account reconciliation. Another project is our TD Online Accounting, which assists small businesses in automating their invoicing [and] receivables.

BAN: When do you decide to automate a system or process?

Margarites: Our decisions around automation focus on client experience and ease of doing business with the bank. We typically target automating processes, but that might mean integration into client systems. For example, if certain client processes around treasury management are manual in nature, we look to help provide our clients with the tools to automate those processes. Whether it’s traditional direct transmission integration, or externalizing our APIs, we invest in making sure our clients can better automate their processes.

BAN: What are some tech challenges you encountered in a recent project, and how did you overcome them?

Margarites: Some of the challenges I’ve seen recently are around automating processes for clients that each have their own unique process. Instead of trying to create a series of custom solutions, we decided to find the point of commonality and build from there. This way, we sat near the end of the client’s workflow and automated everything end-to-end.

BAN: What part do artificial intelligence (AI) and machine learning (ML) play in your approach to technology?

Margarites: AI and machine learning capabilities are a target state for much of our work with technology. The key to building AI/ML-based products is to have a large and accurate set of data; therefore, we pay close attention to how our new products and services align with our data strategy so that we are creating a consistent set of data to enable us to leverage AI and machine learning.

BAN: Can you share some best practices for incorporating novel or unique technologies into existing tech stacks?

Margarites: In my experience, it’s often best to design novel solutions to fit the constraints of your existing tech stack. Making significant changes to an existing stack to incorporate a single solution [or] enhancement can be a significant effort that doesn’t necessarily align with the strategy vision. I find that a careful eye to design and significant client engagement in defining the solution can help ensure the solution operates within the existing structure.

Bank Automation Summit Fall 2022, taking place Sept. 19-20 in Seattle, is a crucial event on automation and automation technology in banking. Learn more and register for Bank Automation Summit Fall 2022.