The creative value of automation with James Briggs

Automation doesn’t take people’s jobs. It helps people focus on the more creative bits of their jobs,…

Julien Bec
February 10, 2021
Scroll to read the aricle

The creative value of automation with James Briggs

Automation doesn’t take people’s jobs. It helps people focus on the more creative bits of their jobs, rather than boring, repetitive stuff.

James Briggs is a data scientist working in automation change processes at UBS. We first met him through his Medium blog, where James publishes weekly articles on a range of topics under the data science umbrella, such as machine learning, NLP, Python, and more.

James has extensive work experience in robotic process automation (RPA) and cognitive automation initiatives, ranging from NLP transformer models to automating technology functions and validation of large-scale quantitative models.

We had a chat with him to talk about his current and past automation projects, his view on the impact of automation in the fintech sector, and his thoughts on what’s in store for automation in the near future.

In fintech, language is having its heyday at the moment. Everyone is really investing a lot of time and effort into leveraging NLP for automation purposes.

James: The most interesting ones that I’ve worked on have to do with making people’s lives at work easier while allowing them to focus on the more creative side of things.

One of my earliest projects involved building a tool that used NLP to automatically identify the most relevant emails amongst millions of emails. This task was performed manually by auditors who really quite despised doing it — and for good reason! This project gave them more time to focus on the creative tasks.

The second example involved cash flow modeling for IFRS-9. This one did not include any AI or NLP, but it was still fairly complex. Business users would spend a lot of time cleaning their data and making up calculations in Excel, without any consistency across the whole company. We helped them standardize their data quality control process and calculations using some Python scripts. This cut down dramatically the time they had to spend on quality control and modeling.

Finally, the automation projects that I’m working on now use robotic process automation (RPA) to automate some repetitive, menial tasks, and NLP to help users quickly find what they need in a big data archive, without having to manually open and scan tons of documents.

Julien: What were the main incentives for automating these tasks and what added value did they bring?

James: Bureaucracy keeps increasing and many tasks are inefficient and end up taking a very long time. Even though digitalization should speed up some processes, there’s still a lot of documentation, repetitive tasks, and policies that slow everything down. Removing these bloated things that people do not want to deal with or that take a long time was definitely a big incentive for these projects — and the low-hanging fruit for automation. As a result, people could focus on the more creative aspects of their jobs.

The other aspect to consider is that some of these projects, such as the cash flow model and the ones that use NLP and AI, go beyond optimizing people’s work as they improve the quality of their work by giving them access to additional data and information in a timely manner. For example, the latest NLP tool we’ve developed works like some sort of a recommendation engine: people type in their question, which has to do with sourcing information from an archive of files, and the tool gives them back a list of the 10 to 20 most relevant sources for what they are looking for. Before this tool was available, people would manually look for information and settle for something that was close enough to what they were looking for as they did not have time to review the entire archive, whereas now they can review a list of the most relevant answers and choose the best one for their case.

Julien: How do people react when processes are automated?

James: Well, it’s definitely not always easy. There are always questions, sometimes fears, and it can be perceived negatively. We usually invest a lot of time in explaining the reasons behind automating a particular process, the challenges automation solves, and the advantages they get out of it. We point out this automation is not there to take their jobs but rather to give them room to focus on smarter and more interesting tasks. Either way, acceptance and understanding of automation tend to be either immediate or not, meaning that people immediately perceive it as a service or as a threat.

Julien: What are the main challenges when it comes to implementing automation in fintech companies?

James: I think there are three main ones. The first one is scoping out the problem, and by that, I mean figuring out the exact process that needs automating. When doing this analysis you sometimes find that you won’t need to automate anything because some steps are just unnecessary. If you do find out that there is a process to automate, you then have to figure out the best way to do it, which is tailored to the company and to the team.

The second big challenge that is specific to the financial sector is that there’s a lot of legacy systems and policies, checks, and controls that were put into place 10 or 20 years ago without the vision of automating or changing them in any way in the future.

The final challenge is the one that is common to any automation process — is it even technically possible?

Julien: What roles does Python play in automation?

James: Python is very well built for automating tasks. You have a lot of existing libraries and frameworks that can help you, and all the data science and machine learning packages are second to none, really. Obviously, there are other tools that you can use for automation projects, which are more user friendly with drag-and-drop interfaces — but these are generally much less flexible than Python. That’s why Python is my go-to for automation.

Julien: How do you explain that, despite all the technological innovations we’ve made, Excel is still prominent in that field?

James: Excel is a very practical tool. It doesn’t take too long to learn how to use the basics, or to start implementing more complex tasks using built-in functions or VBA. And with that, you can already automate a lot of different models. There are still some limitations when it comes to big data or standardizing automation across a whole department, but for the next 10 years at least, Excel will still be there. In order to stop using Excel, you would need to retrain the majority of business users to use different tools.

Julien: Finally, what do you think are the automation trends for the next 5 years?

James: Wow, big question! In the fintech sector specifically, I think that NLP-based automation will become prevalent for all the reasons we’ve talked about so far. More generally, in office-like environments, automation is developing fast across all sectors, addressing repetitive tasks and processes that can be optimized with NLP.

Overall, I see automation as helping people focus more on the creative side of their work and their lives, rather than the boring bits. That’s the creative value of automation!

For more updates on automation, machine learning, NLP & Python, follow James Briggs!

Join our Community

Join our thriving article community and start sharing your insights today!

Like this post? Please share

Latest Articles

View all

What is DirectQuery?

Combine Atoti and Data Warehouses to enjoy the deep-diving capability of Atoti on top of your existing...

Querying with Atoti

In Atoti, we integrate data from different sources into an in-memory data cube. Without worrying about the...

Measuring Interest Rate Risk in the Banking Book (IRRBB) with Atoti

Manage and analyze large amounts of transactions and historical data Interest Rate Risk in the Banking Book...
Follow Us

Atoti Free Community Edition is developed and brought to you by ActiveViam. Learn more about ActiveViam at

Follow Us