In his role, Liberty IT solutions engineer Chris Brennan has used automation to reduce wasteful touches on vehicle insurance claim documents.
Automation can be used across virtually every industry to make a variety of tasks more efficient. For example, Liberty IT solutions engineer Chris Brennan has tapped into automation to reduce wasteful touches on vehicle insurance claim documents.
Brennan has worked in the emerging business and technology division at Liberty IT for the past couple of years, focusing on intelligent automation and extraction.
He told SiliconRepublic.com about what his role entails and what a typical day looks like for him. “If I had to try and describe it, I would say it’s a blend of software architecture design (30pc), software engineering (40pc) and leadership (30pc), all with a twist of innovation!”
He said he could be doing anything from writing thousands of lines of code in a week, designing new solutions on a whiteboard, or conducting research into new tools.
“I might be developing pilots, prototypes and proof-of-concepts to prove how we can further improve business efficiency, or I could simply be on back-to-back calls with Amazon or another cloud provider to request new features be added to the service – it really is that varied and interesting!”
‘I love when we can find a use case for a new technology that makes a tangible impact’
– CHRIS BRENNAN
What types of automation projects do you work on?
It varies, but if I had to pick one stellar project, I would say it’s our AI-enabled Auto Physical Damage (APD) Supplements project.
In this project, the aim was to reduce wasteful touches on vehicle insurance claim documents due to manual triage and upload, by changing ingestion of supplemental information to a controlled ‘pull’ from our customers via a guided, easy-to-use web portal.
Each customer’s transaction flows through a reverse proxy (for security) and interacts with our very own configurable RESTful web service on AWS, which facilitates the ingestion of their data into our back-end. This enables extraction and comprehension of their data across multiple document formats, providing a new set of supplemental data points for further interrogation and intelligent automation.
These data points then flow through a feature engineering pipeline and are utilised as part of a feature set for our custom smart review AI model for auto approval, enabling straight-through processing of claims for our customers.
What skills do you use on a daily basis?
The skills I use most frequently are coding and collaboration with the scaled agile framework (SAFe) at the core of what we do. I work through each program increment with other cross-functional, decoupled, highly efficient agile teams, all working toward a common goal within the same agile release train.
What this means is that it’s imperative for me to collaborate both internally and with technical architects across the wider organisation to design aligned solutions for the enterprise. This is a skill I often draw on more than I originally expected in my current role as a technology lead, but I’m delighted with the knowledge I’m gaining by doing so.
Sometimes this sees that I draw tactical solution designs to rapidly prove out concepts, and at other times it requires that I collaborate on larger design patterns across the organisation in more detail.
Typically, after we’ve agreed a solution design, I’ll lead the technical development of that solution. Most often, I am writing code in Python, JavaScript/TypeScript, and deploying to AWS using their Cloud Development Kit as an infrastructure-as-code technique.
These solutions are often highly complex, advanced, serverless pipelines that could be considered full-stack. They often incorporate a front-end environment in React or Angular that sends data via a reverse proxy in NestJS or Express to an API gateway on AWS, which daisy-chains multiple CloudFormation stacks together, orchestrated through a combination of Step Functions, Event Bridge and Apache Kafka, and of course logging to both Splunk and Datadog.
What are the hardest parts of working in automation?
I’d say the most challenging part of working in automation is managing the inflated expectations about technology. These are often derived from vendors overselling their products and over simplifying complex problems, which can lead our stakeholders into false pretences.
That said, we have robust ways of evaluating these products to ensure they are fit for purpose and can deliver the expected results. Often, we benchmark vendor tools with our own custom in-house approaches, allowing us to choose and/or develop the best technology or approach for the use case – and not just what’s readily available to us.
While this may be challenging at times, it’s this mentality that allows us to push the boundaries of what the technology can do as we work at the intersection of business, technology and value.
Do you have any productivity tips that help you through the day?
Yes, I have an important tip. Protect your time!
I do this by using Microsoft Insights, which uses the data captured from my activity on email and on my calendar to suggest the most optimal blocks of focus time for me. What this means in practice is that I have three hours blocked out every morning to focus, which I normally use as dev time.
But it doesn’t need to be anything as complicated as this. Just do whatever you can do to ensure you work as productively as possible. This can be small things like replying tentatively to a meeting invite and letting the host know that you are available ad hoc instead, or it can be dropping from a meeting early when your part is done. Work smarter not harder.
How has this role changed as the automation sector has evolved?
Automation in insurtech is no longer just about creating tactical solutions to rapidly deliver efficiency savings. Changing customer and stakeholder expectations, along with the emergence of new paradigms such as machine learning operations and hyperautomation, have seen the sector evolve dramatically.
Instead of attempting to automate the interaction with legacy systems to free up human analysts, it’s more about transforming and changing the entire business process by redesigning and reengineering them in data-centric way.
What this means for the role is that we are now developing new applications for the enterprise using the latest technologies. We went from unreliably extracting customer data from free-form text in emails and storing it in systems of record, to developing completely new web applications for our customers to use.
They can now give us their data in one single structured transaction and it’s because of this kind of pattern and data structure that we can now reliably extract, manipulate and store their data in ways which lend themselves to feature engineering and AI.
What do you enjoy most about working in automation?
There are two things I really enjoy. The innovation through the constant emergence of new technologies and the autonomy and flexibility we have in which to adopt them. I love learning about them and I love when we can find a use case for a new technology that makes a tangible impact and delivers business value.
But what I think I might love more than that is the satisfaction I get from this type of work. It’s our insurance that genuinely helps people, sometimes to buy a house or a car, other times for medical care and even to put things back together as quickly as possible when things aren’t going so well for them.
So, working in this area, when something we have designed and developed can have a massive impact on somebody’s life, well that’s just incredible and makes it all the more worthwhile.
What advice would you give to someone who wants to work in automation?
My advice would be to try not to get caught up in the exponential growth of robotic process automation (RPA) and to understand that hyperautomation, while a new term, pre-dates the modern RPA era.
Conventional software engineering tools and techniques and cloud-native, serverless architectures will often achieve far beyond the scope of somewhat sellotaped-together solutions with greater efficiency, reliability and at a reduced cost.
So, when targeting automation, think about what’s the best way to achieve what you want. Is it an approach that interacts with legacy systems and inherits their shortcomings, or would it be better to reengineer and redefine the process? And when answering this question, think about the tools, techniques and skills you would need to achieve it and then focus on building that skillset.
10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.
