Achieving process excellence with AI: a financial services case study

Anna00:10

We're just waiting for Charlie and Gerard, our next speakers, to join. Hello, Charlie. Hello, Gerard.

Anna00:30

We're starting our next session, where we'll discover how to achieve process excellence with AI through a financial services case study. I'm delighted to introduce our speakers: Charlie McGillian, customer success manager at FlowForma, and Gerard Newman, chief technology officer at FlowForma. Welcome, both.

Gerard Newman00:53

Thanks, Anna. It's great to be here today.

Gerard Newman01:07

As we kick off, I'd like to give you a quick introduction to FlowForma. We're all about process automation: we provide a solution that enables both business and IT users to automate processes at scale. We've been doing this for a number of years, and our product contains all the elements you need to automate processes, such as data capture, workflow, document generation, communications, integration and data insights.

Gerard Newman01:42

More recently we've added AI capabilities to the product. Today we'll talk about a case study that highlights some of those AI capabilities and how one of our customers used them. The AI features we've added are practical, and designed to accelerate your process journey and improve process adoption and automation in your organization. First, Charlie will talk us through the case study, and then I'll come back and show you some of the AI features and how this customer used them.

Charlie McGillian02:19

Thank you, Gerard. AI is making a real difference for customers, and today I'm excited to share a case study from one of our clients who has made great progress using AI to accelerate process automation. Unfortunately, they're a bit media shy, so the case study is anonymized.

Charlie McGillian02:56

A quick overview of the customer: they're a wealth management firm serving high-net-worth individuals and investors, with offices across the UK and Ireland. They specialize in high-trust services, investment advice and retirement planning. Now let's look at some of their challenges.

Charlie McGillian03:23

The customer came to FlowForma when they'd hit breaking point following 18 months of solid growth. The new customers they were taking on were creating cracks in their operations, and they were still heavily reliant on manual processes, which was a definite risk.

Charlie McGillian03:45

Their MD and systems analyst reached out to us because their exec team was looking for new ways to modernize. They were drowning in bottlenecks, email chains and disconnected systems, and that was having a big impact on their customers and employees. Now I'll hand over to Anna for a quick poll.

Gerard Newman04:11

Very good. You can see that onboarding is up there as one of the key processes, along with general administration. We see a lot of our customers using FlowForma for onboarding processes, whether it's employee onboarding, customer onboarding or something else, and we think AI can make a big difference there. It helps answer the question of exactly what's involved in onboarding for each use case, and it accelerates building those processes.

Gerard Newman04:48

Approvals, compliance and procurement are also very typical areas. Compliance continues to grow, and structured processes give you the backup information you need to show you're compliant and that standard rules, processes and procedures are followed. So, very interesting.

Gerard Newman05:12

I'll explain how this customer went about automating their processes and meeting their challenges. As Charlie said, they were at breaking point, with a lot of business challenges as they grew. That's not uncommon in organizations that grow rapidly, because processes often don't evolve at the same speed as the organization.

Gerard Newman05:40

They were quite ambitious and wanted to automate a lot of processes as quickly as possible. The alternatives were to increase costs by hiring more people, or to let their process delivery decline, and they didn't want either. So they established three principles. The first was an iterative approach. The old adage that perfection is the enemy of progress comes into play: they wanted processes that were good enough to launch internally, then collect feedback and use the data collected to improve them further. That meant rapid prototyping, getting a process to good enough, getting it out there, getting feedback and improving it.

Gerard Newman06:38

The second was that it would be business-led. The subject matter expertise sat with the core business, but they didn't want all of those people tied up building processes, so they aligned IT resources with them. The IT people worked closely with the business to develop prototypes for them to review.

Gerard Newman07:10

The third key decision was that it would be AI first. They wanted to use AI to save time wherever they could. They didn't have a core competency in process automation, and they felt AI could fill that gap and accelerate their automation journey. They also wanted to use AI to completely automate some of the more mundane tasks people were doing. They'd tried that in the past without much success, but they saw AI as able to take over those tasks, which frees up the team to do more valuable work.

Gerard Newman07:54

Here are a few screenshots of what they were doing. In this one, they're using our Copilot on one of their processes, a financial fact find. With Copilot, they can simply interact with generative AI and ask it questions to find out the key things in a process. In this case, they asked it to build out a financial fact-find process. It identifies the steps that might be in the process and the data points you might want to collect at each step.

Gerard Newman08:38

They could continue to interact with Copilot, or at this stage simply build the process. Building it takes the information provided in Copilot and creates a sequence of steps, which is really the process. Behind the scenes it also creates the rules needed to automate the process and the data points to collect at each step.

Gerard Newman09:12

They could then iterate and improve the process until they were happy that it represented what they needed from a financial fact-finding point of view. Once they were happy, they could launch it to their team and start using it.

Gerard Newman09:37

Here's another example. They'd already been through some process mapping exercises, so they had a map of an existing process, and they wanted to use it to accelerate building a new one. You can see us loading a sketch of a process. In this case it's a hand-drawn sketch, but it could be a photo of a whiteboard, a Visio diagram or any structured process diagram.

Gerard Newman10:25

Copilot interprets the drawing or diagram, identifies the steps it depicts and their sequence, and adds value by identifying the data points that might be collected at each step. These aren't just pieces of text. They can be quite complex data structures: for example, it can identify that a matrix of data is needed at a step and build that out. So they could go from a diagram of a process they'd already mapped to something operational very quickly.

Gerard Newman11:07

When you're working closely with business people, they found that this really brought the process to life. Mapping processes on a whiteboard and going back and forth through iterations can feel theoretical, but when you can build a process this quickly, the theory turns into practice and you get much better feedback and interaction from the user group as they see how their process will look. In this brief demonstration we've gone from a diagram to a working process in a few minutes. They would then refine it further for their organization, perhaps adding more data points or rules.

Gerard Newman12:01

The next thing they were keen on, as I mentioned at the start, was automating some of the tasks in their processes. For that they used a capability we call agents. It's AI-driven, and it's simply a way of instructing an AI agent to carry out a task. We designed it for fairly straightforward tasks that are carried out repeatedly, in high volumes.

Gerard Newman12:38

In this case, we're asking it to extract a passport number from an image of a passport. It's a very common use case, and a lot of customers use it, because it really simplifies data extraction and collection. The passport image is loaded as part of their KYC checks, and we instruct the agent to extract the passport number and return just the number, with no other text or information.

Gerard Newman13:13

Previously, somebody would look at the image, find the passport number and type it into a form. This automates that task. We still have a human in the loop to validate the information, but it significantly speeds up the rate at which these forms can be processed. Charlie, I'll hand back to you to talk about the key wins.

Charlie McGillian13:47

Thanks, Gerard. What I always say to our customers is that they should automate for outcomes, which is exactly what this customer did very well. They had a clear focus on what they wanted to achieve, we agreed their KPIs, and they onboarded to our platform. There's a great quote here from their MD.

Charlie McGillian14:09

Looking at the numbers, they cut their approval times by 65%, which freed up their admin and let them scale with FlowForma without the added stress. They digitized 30 processes in under six months, and it has saved them thousands of hours in annual process admin. We're going to pop into a second poll now, so I'll hand over to you, Anna.

Charlie McGillian14:50

Thank you, Anna. Today we shared a financial services case study, but I know there are a lot of people on this webinar from healthcare, construction, manufacturing and other industries. FlowForma's process automation platform is used across all sectors with great success.

Charlie McGillian15:12

Our customers regularly see a snowball effect: once people see the results from automating a few processes, the news travels fast across the business, ideas come to the forefront, and it's only a matter of time before a lot of processes are digitized across the business. I'll leave this slide up for a moment to give you an idea of what can be achieved across different areas of your business.

Charlie McGillian16:00

That brings us to the end of the presentation. Thank you, Anna.

Anna16:08

Thank you so much, Charlie and Gerard. This session is live, so if you have any questions, please submit them in the Q&A box. Let's start with the questions. We have a backlog of processes to automate, and Copilot seems like it would be very useful for this. Any tips?

Charlie McGillian16:50

I'll take that one. Copilot uses natural language prompts, so everybody across the business can use it. You don't need an IT background; it's a question of using natural language and getting started. It helps you define a process and builds the bones of it for you, and then you can refine that process and add questions and rules. The key starting point is finding the areas you want to digitize.

Gerard Newman17:34

Another point: Copilot uses the large knowledge base behind it, with large language models and generative AI, so it has access to a huge amount of information. For organizations looking for inspiration or a way to start automating particular processes, it can be a real accelerator, and it removes blockers. You can quickly build your first process. It won't be perfect, but you can engage with the organization and tune it. If you have a big backlog and don't have all those processes defined and designed, Copilot is a great place to start.

Anna18:19

How does it differ from options like Power Automate in Office 365?

Gerard Newman18:28

From our point of view, FlowForma is very specifically focused on process automation. It's not a general application build tool in the same way Power Automate is. It's a one-stop shop with all the tools and capabilities needed to automate processes together. The main differentiator is the process focus. FlowForma has focused on this area for many years, and our product has evolved to automate processes effectively, with all the capabilities you'd expect from a process automation or workflow solution.

Anna19:17

Is there a mobile app for FlowForma?

Gerard Newman19:23

Yes, there is. It's used by quite a few of our customers, particularly people who take part in processes while they're out and about. It lets you participate in a process and complete forms on your mobile device while you're away from the office.

Anna19:43

How does FlowForma support compliance and regulation?

Gerard Newman19:49

There's a lot that could be said, but generally, as a process automation environment, FlowForma defines structured processes that must pass certain criteria to progress from one step to the next, which is key in a compliance and regulation environment. That might mean certain information must be present, or that other people must approve steps. In the background, it collects a huge amount of information about the process, the data collected in it and the people who took part, and through our Insights capability you have full access to that data to track and trace everything that happened.

Anna20:41

Another question, from Bill: what happens to the data once it's consumed in the AI discovery process?

Gerard Newman20:52

Nothing happens to it. We don't use the data for any further training. It's used to interact with OpenAI, which we use in the background, to generate a response. Once the response is generated, it's available to you, and you build out your process from it. From then on, that data is part of your process definition, and it remains in your Microsoft 365 environment.

Anna21:29

What types of process diagrams or forms can I upload? Are specific formats supported?

Gerard Newman21:39

This evolves a little, but at the moment we support PDF, which is very common, and most image formats. Between those, we can handle a wide variety of process diagrams: photographs, diagrams created in tools like Visio, or sketches on a piece of paper.

Anna22:09

Another comment: love the case study. Who's the best person to drive AI automation in a business?

Gerard Newman22:20

From our point of view, the ideal team is a hybrid: business people with expertise in the process you're automating, combined with some IT resource, not so much to build the process as to handle any integration that's needed. FlowForma has always been designed as a no-code product, and we encourage business users to automate processes themselves. AI enhances that further, because you're building your process with natural language prompts, which is quite easy to do.

Anna23:09

There's a question about onboarding. What does onboarding to the product look like, and is much training needed on these AI features?

Charlie McGillian23:22

I can take that, as it's part of my team. The onboarding process is pretty straightforward. We work with the client on timelines and set expectations, and we aim to have a client live and up and running within six weeks. With the AI features, that's now a lot sooner, probably within a few weeks.

Anna23:56

How does it support audit trails and process transparency?

Gerard Newman24:12

I covered some of that on the compliance side. The main thing is collecting information about everything that happens in the process, and we see two sets of data. There's the operational side: what happened, where, when and who did it. Then there's the business data collected, for example, a list of assets and their values in our financial fact-find case. Collecting all that data and making it available in our Insights module gives you a full audit trail, so you can drill down from top to bottom and understand exactly what happened at each step.

Anna24:57

A question from Kier: who owns the process tools once they're in place? If we start with FlowForma, are we tied in forever?

Gerard Newman25:10

Each process tool has its own distinct capabilities. FlowForma sits within your Microsoft 365 environment, and building processes and interacting with Copilot are carried out by your own organization. The processes are then implemented and run within the FlowForma application.

Anna25:37

We have two minutes left, and a final question: can FlowForma identify and list the different risks within a process that's to be automated?

Gerard Newman25:50

Yes, that's the kind of interaction you get with Copilot. If you're looking at specific risks or areas that might need tightening up, you can keep interacting with Copilot and ask what the weak areas of the process are, or where you might need more controls. People really like being able to have a conversation about their process until they're happy enough to build it and have something they can operate with.

Anna26:31

Brilliant. Thank you so much, Charlie and Gerard, that was a really interesting presentation. Before we finish, would you like to say anything to our viewers?

Gerard Newman26:42

Just thanks very much for your time. It's a very exciting time in the process space, with AI and everything we can do with it, and I think it helps overcome some of the process blockers people have had in the past.

Charlie McGillian26:58

And thank you very much for taking the time to be with us today. We look forward to speaking with you all in the future.