How AEGIS Insurance Automated its Underwriting Workflow

Shirley Shea, an insurance systems consultant and former IT executive for AIG, has spent the past 6 years working with AEGIS Insurance Services upgrading the specialty insurer’s aging infrastructure. She recently sat down with Insurance Networking News to talk about the sweeping overhaul and why AEGIS, which serves the utility and energy industries, started with underwriting and moved from the front office to the back, as it deployed Pegasystems throughout the entire company.

INN: How has the policy administration system at AEGIS changed and evolved over the past six years and why? What sorts of market pressures has the company been responding to?

Shea: AEGIS is a privately held mutual company. so and they need to be able to respond quickly to brokerages that are bringing in business from the various member companies. They also need to be able to quickly rate a wide variety of different policy types and get them out the door. When we began this project, this took a couple of weeks at a minimum. The underwriting and the workflows were manual, which was slowing the whole process down. There was limited underwriting automation, yet multiple functions within the company participate in the process, including a loss control group. So the most important thing was to make the system more efficient and able to respond more quickly.

See also: PAS Transformation: A Well-Calculated Risk

INN: Tell us about the workflow automation; to streamline and automate all these processes must have required some pretty dramatic changes.

Shea: One of the things we did for underwriting was to develop a customer-facing online app to help acquire more information about the customer and risks involved. This is an automated process that collects data and gathers it into the back end for AEGIS’ underwriters to review and process.

Underwriters take that information and put it into the new system, and different underwriters and companies have different levels of authority. They’re issued a contract that says, without any other approval, they can issue policies with risks up to 'X,' and retentions up to 'Y.' Rules for when an underwriter needs approval or verification from a superior, or someone in another division, are all built into that system.

Policies that require approval, along with a notification, are automatically sent to the appropriate parties for review. They have their own review screens and methodology, and can approve or reject and indicate why. It’s all done in Pegasystems.

INN: Is that a matter of configuring it, or programming it?

Shea: We stayed within what Pega called the guardrails, but we wrote a little Java script ourselves. It was mainly to enhance the user interface. But all of the workflows are out of the box and configured through the Pega system.

INN: What does the interface look like?

Shea: Pega is really good at intent-driven processing, so if you do ‘A’ here, or ‘B’ there, they automatically take you through a path, based on your configuration. But the underwriters also wanted to break out of that flow and have their processing and navigation driven by menus in the screens. So unlike some of the Pega screens that you might see, we have a menu bar across the top that’s specific to AEGIS, that we built ourselves. When we upgrade the underwriting system we’ll be replacing that with as much Pega out-of-the box as we can. That’s what makes our development go so quickly. For instance, we do some Canadian and out-of-the-United-States type business, and we have ‘hovers’ that convert; when you enter Canadian dollars you can hover over the number and the system automatically converts it back to U.S. dollars, so you know what all the equivalents are. We have a lot of help text, you know, depending on what the users felt the need for.

INN: What systems did you integrate with Pega?

Shea: Our system integrates with a very old Siebel CRMsystem, which was configured by AEGIS. It houses all of the member, broker and brokerage information, and all of the contacts, etc. We interface with that directly to put the brokers on the policies themselves. We also integrate with our current invoicing system, which is a very old Great Plains system. We do integrations to that directly and then we use SSIS to reformat the data and get it into the warehouse.

INN: AEGIS had tried this before. What did you do differently this time around and how long did it take?

Shea: The system they bought previously wasn’t meeting 80 or 90 percent of their needs. They basically were trying to rewrite the system they had just bought. And the other thing was, they were trying to replace everything downstream at the same time, We completed coding and testing within the six months and went live in nine. The design method we used was to actually persist the data externally to the underwriting system in the same format that the other systems were used to, so that all of the feeds didn’t need to be rewritten from scratch. Once we got that up and running, we replaced each one of those.

INN: How many people worked on the project?

Shea:  We had an on-shore, off-shore agreement with a consulting company in India. Offshore, for the first phase of underwriting, we had maybe six programmers. On shore, we had one lead business analyst, one architect, and one business analyst We also had two folks from Pega’s  professional services team, because we basically wanted a blessing on what we did the first time around. I had used Pega in my past life. But I also know, from the 28 years of managing IT at AIG, that maintaining the scope of something that’s easy to configure is a challenge.

INN: Knowing what you know now, what would you do differently?

Shea: I would probably break up the deliverables into smaller pieces and work interactively with the business team a little more often. The hard part about building a new system and making it work is the integration: the outputs for the policies, the reporting pieces of it. You have to consider where your reports are going to come from, how are you going to feed your data warehouse, how are you going to get that policy back to the business people. You can do all you want internally, but if you can’t get the policy out the door, you really haven’t done your job. So it’s the stuff around the system that always catches you. It’s the detail around it that really gets you in trouble.

shirley-shae.jpg

For reprint and licensing requests for this article, click here.
Data and information management Workforce management Analytics
MORE FROM DIGITAL INSURANCE