We no longer support this browser. Using a supported browser will provide a better experience.

Please update your browser.

Close browser message

Technology

at Our Firm

How Intelligent Automation Integrates into Bank Processes at J.P. Morgan

Intelligent Automation is a term used often in the tech space, but how does it integrate with AI and machine learning to produce business benefits? In this episode Anish Bhimani is joined by J.P. Morgan’s Shefali Shah for a conversation on how Intelligent Automation can increase organizational efficiency, free up human intellect and enhance customer service. If there’s a specific topic you’d like us to cover, email us at: tech.trends@jpmchase.com.

Play button

[electronic music]

Anish Bhimani:

Welcome to TechTrends. TechTrends is a podcast series that provides perspective and latest trends in technology, FinTech and digital. On today's episode, we're going to provide some insights into intelligent automation, what it is, how it works, and how it’s becoming an imperative for organizations that want to reach higher levels of optimization in efficiency. I'm Anish Bhimani, Chief Information Officer for Commercial Banking. And joining me today is Shefali Shah, head of intelligent automation for the corporate investment bank at JPMorgan Chase. Shefali, welcome to TechTrends.

Shefali Shah:

Thank you, Anish, great to be here.

Anish Bhimani:

So Shefali, we hear a lot about intelligent automation. It's a term that's used often in the context of things like robotics or bots or RPA. Can you describe for our viewers what intelligent automation is and how long it's been a focus of study?

Shefali Shah:

Sure. Intelligent automation actually is a spectrum of software and technology that allows you to mimic human activity. It is about manual and ineffective work done by human in an office environment. What we know it is a white collar work and it applies to all industries and organization. The spectrum of tool includes basic software that mimics humans activity of typing, searching, downloading, and such to more advanced level activity around data transformation, data digitization, documented digitization. And even some of we have seen in our personal life chat bots that allows us to automate conversation between humans. It is the latest trend that's evolving very quickly. This automation in this solution helps an organization improve their efficiency, free up their intellect, human capital, to do more advanced level work. And it also enhances customer service in a business benefit.

Anish Bhimani:

So one of the other terms, we hear a lot. We talked about RPA, which is robotic process automation. People talk about robotics in this context. And a lot of people hear robotics and they think about physical robots and manufacturing plants and things like that. How does RPA differ from sort of traditional robotics in that sense? Can you explain what RPA is as well?

Shefali Shah:

Sure. Yeah, it is not a physical bot running around your office or taking seats. It's actually a software that works noninvasively on our desktop or on top of our technology that we use say if it's Excel or your finance application or in our case processing application. So it's a non-invasive software that allows you to mimic our activity of typing. So it's definitely not physical bots. And that was a revolution we saw with manufacturing and farming. I think this is a new revolution around digitizing our work that we do manually on your computer.

Anish Bhimani:

So thanks for breaking that down for us. Can you talk a little bit about your role and your team and how you focus on integrating intelligent automation into the day-to-day processes at the bank?

Shefali Shah:

Sure, yeah. I've been leading this organization of intelligent automation services for past three years. And part of my role is to bring in this latest technology through strategy and research and looking at what's coming up in industry to the firm for corporate investment bank. I also sit at a firm wide council, so collaborate with my colleagues and peers across the firm. What we've been doing for past three years, that we started out with a basic bot solution to take away some of our manual work, especially in our global services center.

And we have evolved from there to integrating data transformation, data integration, our technology, and also now partnering with our machine learning and AI team to incorporate more advanced level of technology like NLP to automate unstructured data into more structured data. Other thing we do in my team is to partner with our operations manager and the teams doing the work to define best practices and deploy these solutions in a more controlled manner. So we centrally govern and manage all the initiatives to make sure they're getting done in an effective and an optimal manner.

Anish Bhimani:

So can you talk about some of the benefits of intelligent automation? Is this really more of a cost savings effort? Is it more of a...xWhy would companies be interested in doing this?

Shefali Shah:

Sure, absolutely. So some of the areas in corporate investment banking where we have deployed this solution over the last three, four years is around account set up, transaction processing, reconciliation and also inquiry management and generating metrics and reporting. So if you hear some of these words, this tends to be very manual process. So in this case, allowing automation or RPA to automate these processes has freed up our teams to focus more on client service. We have scaled up our efficiency as well to free up the team to focus on more value add activity. And also this has been a great interim solution for us as we invest and digitize our old

platforms into new platforms. So I think the solutions does provide us some quick wins and some immediate relief to our teams to free up their time to add to a value add activity.

Anish Bhimani:

So to summarize really it's about automating the easy stuff so that the people can focus on the harder problems. Focus on more intellectually difficult or intellectually challenging kind of work that might even be more interesting for them and helping them sort of balance their workload and lighten some of the load on the operations teams. Is that it?

Shefali Shah:

Yes, that's part of it. Another part of this is also up-skilling our workforce and our teams to do more advanced level of technical work, exception resolution and creative problem solving and root cause analysis type of work. So yeah, definitely. It is to free up our human capital to do more value, add intellect work. For sure.

Anish Bhimani:

The other thing that you mentioned is the idea of RPA as an interim step or a quick wins or things like that. Often automation, if you'll call sort of. It's a proper automation, building it into the application, something like that can take a long time. And it can be, "Hey, it's going to take months or years to get this going. Do you see RPA as sort of a way to, okay, let's get this quick win, throw off some capital so that you can invest in the proper automation." Or things like that. Is that it?

Shefali Shah:

Yes, definitely. That is part of it. So we really partner with our technology team and our platform technology so it's not just putting bots randomly on a solution or a process. We really partner with our technology team, operations team to look at what is it worth doing now with RPA and what is it where we can integrate solutions, especially with latest API functionality where you can integrate data. So yeah, it is partnering with the technology team to optimize our end to end processes, or also, not to invest in the technology for some ad hoc one-off processes that can be automated through RPA.

Anish Bhimani:

So in past episodes, we've talked about artificial intelligence and machine learning. Can you talk a little bit about how intelligent automation interacts or relates to AI and ML? Do they compete with each other? Are they complimentary? How do you think about that?

Shefali Shah:

Sure. Great question. It can be a bit confusing for sure. Lots of buzzwords there. So the way to think about it is intelligent automation, RPA, mimics humans activities. So it's what we do. AI is more around how we behave and what our intelligent is. So it kind of reproduces our behavior and ML, machine learning, it learns and does things based on data. So pattern, prediction, and so forth. And without getting into more technical, that's how I like to kind of break those down. And I truly think they compliment and integrate well with each other, for sure. And one example I would give you is in our reconciliation process, where we started out with using RPA to kind of mimic the activity of researching and finding breaks and reason for breaks that our teams were doing manually to now, then we partner with our machine learning team to move that over into identifying based on the pattern that the bot identified to a processing.

We now are able to integrate machine learning to identify patterns and even resolve those exceptions more rapidly. So that's an example of how the two can be integrated. On a more advanced level side, we're using something called natural language processing now in conjunction with RPA to be able to read unstructured data, unstructured document, and text. And also some of our unstructured emails that you may get internally or externally to identify pattern and structure, to then be able to allow the RPA or the bots to process that transaction now that we have a structure information. So that's how I see the pre interacting, but it's an evolving space and we're continuing to learn more about it.

Anish Bhimani:

So to summarize, intelligent automation really mimics human activity and AI and ML mimic human judgment, maybe as a way to think about it, right? But what you're saying is they can be very complimentary. In your example of taking unstructured data, you can use machine learning techniques to feed into RPA or processes like that as well. Yeah?

Shefali Shah:

Correct. Exactly.

Anish Bhimani:

So let's talk about COVID. The COVID pandemic obviously challenged us to think and operate in a lot of different ways. In light of all that, can you talk about how intelligent automation can be used and has been used to help organizations be more resilient?

Shefali Shah:

Yeah, absolutely. Thankfully for us, since we had quite a few bots already running or IA solution already implemented during COVID, we were very thankful that all of those solutions were running as planned, even though all of us were trying to figure out where we were working from and getting our home office set up our IA solutions were up and running with no disruption, which really helped us keep up with the volume and the uncertainty. During crisis, you don't have time to invest in kind of look at things end to end processes, you want to get something done quickly. And especially in our case with market volatility and the volume, going up three, four times as a normal levels within the first week, or few weeks of the crisis, we were able to enhance our solutions or add more bots.

Let's say, our digital workforce to cope up with that volume where we can add human and train them as quickly as we would need it to. A couple of others area during this crisis, this solution helped us is in our consumer bank where we process a lot of our small business loans that we had to do, which was a brand new process that was introduced where we were able to automate those through leveraging IA solutions. And then finally, we also leveraged it for some of our payment relief program that we provided to our consumers as well. So yeah, during crisis, and for those one off ad hoc urgent type of processing requirement, this solution could be great benefit to the organization.

Anish Bhimani:

So it sounds like that intelligent automation can be a tool that can help you sort of rapidly get up to speed on a new process. So things like that scale up, things that you might've had to do before. I think what we have seen during COVID is there's been a lot of that. We've got to get this going up and going in a week and do it at scale and other things like that. Yep?

Shefali Shah:

Yep, absolutely. Yeah.

Anish Bhimani:

Okay. So one of the themes that we talk about a lot is just because you can do something doesn't mean that you should do something. And there are a lot of use cases that I think benefit tremendously from intelligent automation and maybe some others where it's not appropriate. It's been said that situations where you have a human and a machine working together rather than technology alone often delivers the most benefit to an organization. With that said, can you talk about the process for what it's like to implement intelligent automation and what are the functions or operations or roles that would benefit most from this kind of technology?

Shefali Shah:

Yes. Implementation of intelligent automation along with any technology upgrades or transformation revolves around three things. It's people, process and technology. So I think the key to successful implementation of intelligent automation is to first and foremost have a sponsorship and priority from the top. So your business owner, your technology CTOs must prioritize digitization and automation for their organization. And to do that, you need to kind of make that as your key objective and invest in the technology to partner with the right vendor or a solution provider, which there are many out there. So select the one that's best for you. Luckily with this solution, the upfront cost is minimal, not too high, which is manageable. And the return on that investment can be achieved very quickly. So recognizing that and prioritizing that is a key.

 The second is to really engage your process owner and the teams that are processing the manual work to make them understand the value of this digitization and automation and take the fear out of bots are going to take my job away. It is about upskilling their skills and making them more digitally savvy. Everybody in their personal life has smart phones. And we can do lots of things with Alexa and whatnot. I think, in a work environment, we need to introduce this digitization and automation to our teams. So that's a key engagement for the people is another key aspect of it upfront. And the third is to then make sure you set up the center of excellence or a project, or owner of this digitization activity so that there is a central control and best practices and accountability across the team.

And then finally, it's just finding the right solution and vendor. There is no wrong answer with this to look at various solutions that are out in the industry and look at the broad spectrum of solution like we talked about from doing activity of simple bots to more data digitization, document digitization and make sure a solution that provides a broad spectrum of solutions. I think if you integrate people, process and technology and make that as a center, I think you can be very successful at introducing this.

Anish Bhimani:

That's a great way to think about it, I think. So you find the right process that's there, something that is reasonably well-defined. What a lot of people have said is if you can write out the process, then it's a good candidate for RPA or IAA. And if you can't and there's judgment involved, maybe it's not. But I think getting the operations teams or the individuals on board is a very important part because it really does need to be a partnership between the technology team, the operations team and the business process improvement folks.

Shefali Shah:

Yeah. Very true. I think you hit on a good point where there is a balance of not automating a bad process, so really reviewing your end to end process and taking the time to possibly redesign it. And re-imagine it, how it can be done through automation. So that's a key aspect upfront for sure. But there are times also where there's this balance of it may take too long to redesign everything. There are other teams involved or external dependencies on that document you receive or so forth where you can’t change things rapidly. In that case, you need to break down the process and automate what's in your control and then integrate other aspects of the end to end process. So yeah, absolutely engaging people doing the process and really evaluating your process is a very important factor.

Anish Bhimani:

Yeah. That's another great point because if you automate a bad process, all you end up with is a bad automated process. Right?

Shefali Shah:

Exactly.

Anish Bhimani:

So the other thing that's often been said is as soon as you put a bot in place, you want to think about how you get rid of it, as you said, it's a good sort of short-term solution while you can do that longer term automation or model work, or other things like that. You mentioned that this has been a field of study for five to 10 years, I think, as you talked about it before, and how has automation technology evolved over that time? And what are some of the more popular trends we see right now, obviously we've talked a lot about robotics and RPA, but there's a lot of new, exciting technology coming to this space as well, that can do a lot more, right?

Shefali Shah:

Sure. It is evolving technology. And what I see the biggest aspect coming out of this is integration of this various solution into one platform. And one solution that kind of lets you do multiple things depending on your end to end process, not just mimicking it, but incorporating some level of your judgment, and decision-making, so I think integrating some of these solution is becoming a key. Other thing you would hear and I think I'm getting excited about is automating process of automation. We talked about reviewing end to end process. We talked about re-engineering. What's evolving is digital process discovery and data driven decisions, data driven automation. So we're seeing a lot of integration of those aspect of where you can do your process mining, your process evaluation more digitally based on the data that you have or that you behave, watch how your humans are doing the work.

So we've seen a lot of technology coming out that you can record processes as the person is doing it so they don't have to write it down or explain. You kind of observe how they're doing it so that allows us to then take that information and design your automated process more efficiently. The other area we're seeing is definitely integration of this machine learning and a more advanced level of AI, but that I think it goes into more strategic platform as well. The other aspect of it is we talked about, we want quick solution. We want to get up and running quickly. So a lot of what we're seeing is cloud enablement of the solution. You have to invest even less upfront in installing this software in your premises or in your thing, but we need to make sure when you run things in cloud, it's in a controlled manner, you can access it who's doing what and so forth.

The third aspect of it is we're seeing more citizen developer and what's called as an attended bot versus unattended bot. In the past few years, we've deployed a lot of unattended bots where they run in the background, but attended bots are something you can request when you want to do the process. And also the citizen developer concept is coming up where we allow the person actually doing the process rather than relying on the technology team to build a bot or build a solution, the citizen developer who are [inaudible] themselves, even with no technical background with no code, no code solution can build their own solutions. So that's where things are evolving. What I really would like to see is a balance of the end to end process review versus the one-off citizen developer processes. I think both has a value we just need to find the right balance. That's where I'm seeing things evolving.

Anish Bhimani:

There's a lot of interesting themes in there. I want to go back to this concept of process discovery, right? So I think a lot of people, might look at this space and say, "All right, I have to have this well-defined process. And it's an algorithm that can be followed, etc." And I think what you're saying is, "Well, with some of these new technologies, maybe you don't." You can sort of say, "Well, look, I don't know how things are done today, but I can sort of put a recording on the person doing their work and sort of infer what that process is." Which is a great way to get started with this. I think it's important to know to the point that you had mentioned that you still have to then go and look at, is that the right process? Just because that's how it's done today you still want to do your process improvement, things like that, right?

Shefali Shah:

Yeah, absolutely. Yeah. It's all about I say is combining art and a science. So you do still need intellect of your process owner, process redesign analyst to come in and help with this and balance the art and the science together.

Anish Bhimani:

And the other comment I like the way you talk about citizen developers. So with these low-code or no-code platforms, maybe you don't have as much dependency on a technology organization to come in and build new applications or tools or other things like that. How much do you engage the individuals? Let's say the operations teams whose processes you're tending to automate in the solution of that. And then can that be used as a way to sort of drive greater engagement and buy in on the process from the people who might've felt in other times that, "Oh, the robots are coming to take my job." And then they become part of the solution, rather than something being done to them.

Shefali Shah:

Yeah. I think as we look at more and more value of humans, especially knowing that last year, the work-life balance, right? Engaging the people doing the work is very critical and then giving them the new skills, the technical skills and this low code skills and making them feel engaged and empowered. I think it's the best way forward and still keeping your technology team informed and aligned so they know where you're putting this solution and this almost becomes a requirement for your future enhancements as well with your technology team.

So I think engaging individuals doing the work and the teams that are supporting that technology platform is very critical. And I think in the long run, everybody wants a more value added work. I think people are not fearing this technology and digitization. It's only going to create new roles and new jobs, I think in all industries across the globe. So yeah, that engagement is very critical and we're seeing that in our teams where people are feeling a lot more valued by learning new skills, and we're almost mandating them to learn this new skills as they look for career advancements as well.

Anish Bhimani:

That's great. That's great. It's a great combination of evolution and automation of the environment, but also upskilling the team and getting people excited about what the future holds for them.

Shefali Shah:

Yes, absolutely. Yeah. Let's all hope that through this technology and digitization our world evolves, and we free up our time from this manual work and mundane work to doing more value add and free up more time for ourselves and solve more bigger problems that we have in the world today. And it will also, I think help us create equalities and more opportunities. So very exciting times ahead for sure, with this technology.

Anish Bhimani:

Yeah. That's great. Very exciting. Shefali, thanks so much for joining us and for providing your insights on such an exciting topic.

Shefali Shah:

Thank you Anish for having me here. And I hope this information is useful to our viewers.

Anish Bhimani:

And to all of our viewers, thanks very much for joining us today. Remember if you liked this episode, you can subscribe and rate us on Apple Podcast, Spotify, or wherever you get your podcasts. See you next time.

[electronic music]

END

Technology Blog

Our outlook on technology, written by distinguished engineers, machine learning experts and AI researchers.

Publications

Technology Publications

Explore a selection of recent J.P. Morgan publications that are authored by the firm’s machine learning engineers, AI researchers and quant experts across the academic community.

Read More

Collaborate With Us

Perspective

Learn more

J.P. Morgan Developer

Learn more

Bot Marketplace

Learn more

Teams

Meet the teams of data scientists, engineers and researchers

Careers

Technology Careers at J.P. Morgan

Explore our career opportunities to design, build and deploy tomorrow’s innovative technology.

Apply