What is our primary use case?
The primary use case is within anti-money-laundering: To take alerted cases and gather artifacts on multiple websites or applications, and then to consolidate those into a single file, in this case in OneNote, and provide that to the investigation team.
Primarily, this was all being outsourced because it's a lot of very tedious researching and collecting of the artifacts and consolidating them. Once that consolidation is done, those artifacts can be handed off in a structured format to another team which actually looks at the alerted information, at the details, to see whether it justifies the alert. They can make a decision based on the artifacts at that point.
In addition, with the solution going out there and being visible, we've had interest from the insurance side that wants to jump in. They've got several use cases that they would like to do. The capital markets group has use cases they would like to get involved, and the bank itself has use cases that it is just beginning now. It has ramped up a team already and they're going to start wanting to jump in too. Even groups like DevOps and some other groups that are more cost centers are wanting to jump in. Anything they can do to lower their costs helps out the bottom-line dollar for the whole company.
Looking at a company like ours, with so many different departments and processes and policies, and so many manual tasks, the use cases that we can have for both attended and unattended bots are pretty much countless.
How has it helped my organization?
There was a large number of individuals who were doing the data collection and artifacts collection, and they were actually a third-party. The solution improves our situation in terms of time, money, and resources. Plus, when you're dealing with a bank's information, there are additional complications of privacy concerns. If we can keep that in-house and have a bot or automated code take care of it, then there are fewer human eyes on the private information as well. It's a resource saver, it's a money saver, and it helps us with security, keeping more human eyes away from private or touchy information.
What is most valuable?
Its integration with source control is very refreshing. There could be a little bit more maturity around how to do some of the features, but not having to go outside the app, not having a separate set of procedures, allows you to check in and check out right from within the app. You can make edits and uploads and undo your checkouts. That integration is very nice.
It really does have a great dashboard. It has Bot Insight, it has MetaBots, and IQ Bots. It has so many features where it can read a file and loop.
What needs improvement?
Regarding the integration with source control, when there's a large number of operations the bulk operations need a little bit more maturity.
Also, regarding the UI, once you're in some of the screens there are fixed dialogues. They are a set size, so there's a lot of horizontal scrolls and vertical scrolls. Those are things that can be addressed in the future. The integration and having it inside the applications are far more important than these pesky complaints. But I get to a lot of scrollbars when I'm reading code. Sometimes I have to keep scrolling and moving up and down and it's a bit of a nuisance because I'm focusing more on navigating than on the actual logic that I'm trying to read through. A little bit more friendliness in the UI would help.
I came from OpenSpan and Pega and it's a different approach as far as the coding goes. One thing I feel limited in, in Automation Anywhere, is that sometimes they give these wonderful screens where you can do conditional loops or branching with the "If" statements and they'll have these built-in features for if a window exists or if a folder exists or if a process is running. But if I want to do something that is more purely business-related and is somewhat complex to write or is nuanced, I sometimes have trouble implementing that inside of the Automation Anywhere script. I feel limited at times with some of the looping and some of the branching and some of the ways to make procedure calls when I have a complex business issue.
Where it's really great, where it's very simple, is for me to see if a window exists, to see if a process is running or if I'm waiting for a window to close. On the flip side, let's say we're in capital markets and they have a complex business rule for some of the analysis. I don't know how well the product would handle that case. I can't say that it can, I can't say that it can't, I just have concerns of how that would be done, at this point.
For how long have I used the solution?
Less than one year.
What do I think about the stability of the solution?
If we're talking about the Control Room, those services stay up. We have them on a Windows Server, and I haven't seen it have an issue standing up, other than when we have some code that has gone rogue. When doing development and there has been a mistake in the code, we've had to redo the server a couple of times. Given that, I would say the stability is an eight out of ten.
We've had to change our configuration a couple of times. I'm only saying eight because I know there are scenarios that we haven't come up against and when we write code there are going to be times where Windows or the server has to be rebooted.
We did have one issue where it kept disconnecting in development but that appeared to be a VPN issue and we got help from Automation Anywhere on that.
Overall, it's very solid. I don't know that I could give a perfect score to any software out there.
What do I think about the scalability of the solution?
The scalability is limited more by our money and our hardware than anything. The scalability really depends on how much RAM and how much network bandwidth we can do, how many servers we can apply. I know we can just keep adding to the cluster and I know clients could keep popping up. Since we're at the forefront of this, it has not been an issue.
However, I do know that, within a year, when we start having multiple clients running and we have multiple developers in there, I may have a different response. But, again, I think we would just have to add more Control Room servers and more resources to the servers. We haven't hit a scalability limit issue yet.
How are customer service and technical support?
The technical support has been very prompt. They've been very willing to help out and, in almost all cases, they've been able to provide an answer or solution relatively quickly. If anything, we have been the bottleneck because they have been more responsive and quicker than we were able to implement. They may want us to make a change or tweak, give us an answer within an hour, and we have to get permission or approvals to make that change in the environment it's in.
I think that's another huge plus for Automation Anywhere: their customer service and their customer relationship.
Which solution did I use previously and why did I switch?
Our company didn't have any RPA enterprise-capable tools. I don't know what motivated them to learn about it. It may have come down from the executive level saying, "We want RPA, go figure out what it is." That's when they went and looked it up and started researching. They did like their competitive analysis of all the RPA software tools and platforms that were out there. They narrowed that down to ten, then to 5, and then there were three that they ended up with and did a matrix comparison. The matrix included risk, ROI, and cost and they came up with a weighted system. Automation Anywhere came out on top.
How was the initial setup?
The initial setup is a complex process that they make very straightforward. Their installation is very quick. It was relatively painless. Any pain that we experienced was on our side because we had to make sure we had the right permissions or direct privileges, or that we had a firewall that was configured properly. We had a relatively complex site where we had clustered Control Rooms and we're using load-balancing. Within a day or so, we definitely have it standing up and, typically, within the same day, we have it configured. That's assuming we don't come into any complications on our side where we have to go ask for permission to get access to something or for a new certification. Overall, their installation is fantastic.
It took about two days to stand it up, to configure it, and then to smoke-test it, and make it productive.
The setup was prior to my coming aboard, by a couple of months. Our company had talked with IBM and Automation Anywhere, so the strategy was definitely to go through the documentation and to have an Automation Anywhere expert help with the development environment. After having meetings and reading the documentation, they had a hand-holding approach in development. Then they documented the steps. They went into the next environment, ran through their documentation, updating it because there were some changes in the clustering as well and the load balancing. They got that standing up and documented that. By that time, they felt comfortable in production. They were able to go through and repeat the steps without having to go back to Automation Anywhere or IBM for support.
We had to repeat the process here about two months ago and set up a new model and set up a new production environment and some more servers. That's where I get the couple-of-days timeline from. I was following their documentation with our internal guy who did it. Obviously, we had some guys that do database, some guys that do the Windows Servers, and then, myself; I was doing the actual client side. We're all on the phone at the same time taking care of it and it's less than a day.
In terms of the very initial setup, starting from the planning stage, I wasn't here at that time. I would say that it was about four to six weeks, but that's mainly because they had other stuff that was going on, so they were just having the weekly meetings until they got ready to try it. I do know they started talking in May and by the end of July, they had it done, but there were gaps where they weren't working on it in there.
The actual length of time may be about a week because we have to do a change request. We have to go through a procedure where we get approval from the business managers and the lines of business saying, "Okay, we're ready to go live. We're going to go ahead and push this into production," and we need to do backups and have a contingency plan. We then have a meeting and make sure everybody is okay with the current test results. Once all that's done, we can deploy in one night, have it smoke-tested, and have it running the next day.
For the actual deployment steps, you really only need one person but, typically, we will have the developer, a business person to do a smoke test, a Windows person and a database person; four people, only because we have separate roles here. Technically, it only takes one person, but developers don't have any permissions. But we need a developer in case something goes awry to help out the build person who's pushing it. If there's something they can't do, then they need the Windows person to handle any kind of Windows services. And if there's a database issue, you need a database person to run a script. Four would be max, but that's actually very typical in a lot of corporate deployments.
What was our ROI?
We're just getting into production right now and we're handling the first wave of production issues for getting it into production. Unfortunately, we got our code from a vendor that was helping out initially. Even though I've been here six months, I just got my hands on that code a couple of months ago, so we're having to do some cleanup.
We're really hoping by the end of next month or the month after to get a good idea of metrics on what our performance is, how fast we can work, the ROI, and the offset. We're almost there, but it's a little premature to determine ROI.
Regarding areas for ROI with this solution, the first one is that we will be able to terminate a contract for doing all this work that is going to be replaced with the bots. That immediately means lower legal costs, less overhead, less money that needs to be paid out. It's two or three bots that will be replacing multiple people. That's the initial type of ROI we can see.
It's a scale game, as well. The same code that we're doing is very applicable to a couple of other groups within compliance, within audits, which will be collecting very similar type of information from a couple of different apps. The ROI there is going to only increase because we'll have this reusable code that we can extend to other groups very quickly. I think that's why they chose this particular path. So hopefully, we'll be able to scale this ROI tremendously in the next 12 months.
Policies and procedures always change. The question is, can we be nimble enough? Can we build a platform that is solid enough, and have a pool that is talented enough to be able to produce bots in a cheap enough fashion to realize that ROI? So far, I don't see anything that would restrict that or cause that not to happen.
What's my experience with pricing, setup cost, and licensing?
Automation Anywhere's pricing is competitive. That's obviously something that attracted our company to it. They're very well priced. I can't speak to let's say UiPath or Blue Prism. I do know Pega vs Automation Anywhere is somewhat comparable, but Pega also requires a lot more infrastructure and a lot more experience to get up and running. There's a bigger upfront cost to get Pega and they also want to push their case management, so even if you go RPA, they're going to want to get you to do the case management side of things as well.
Which other solutions did I evaluate?
The shortlist was Pega's RPA and UiPath. Blue Prism was in the top-ten but they were not in the shortlist because they never responded to calls. It was felt that if they couldn't respond to the sales calls how could they go forward?
What other advice do I have?
Do your homework, talk with the team, get your questions, read through the documentation, and then decide on your platforms and make sure you really focus in on whether you need clustering and the load-balancing because those are going to make a big difference in your costs, your platform, and scalability. Get that worked out first.
Then pick a use case that is very quick and simple where you don't care about the ROI but what you want to do is make sure that you're testing your environment, that all your environments work, that you can do source control, that you can promote, that you can unit-test, that you can do regular tests, that you can do deployments. You can solve all those problems without the headaches of trying to figure out how to keep the business happy, how to keep the cost down, and just focus on making your environment solid.
Hopefully, that use case is something small enough that you can do within a month or two. Once it's deployed you can see how to support it, how you test it. Then you have time to focus on your standards. What are your programming standards? What are your deployment standards? What are your guidelines for coming up with change requests? Those things, ultimately, regardless of the code, are always going to be your success and failure points.
Bottom line, when you get down to it: The coding is not going to be the bottleneck anymore, it's going to be your procedures and policies around it, your project management. Focus on that.
Automation Anywhere has made it very easy for you to install, they've got the tools to make it very simple for you to create a quick, small application and to get out there. Again, forget the ROI the first time. Get it working, get everything panned out, and then, once you feel comfortable, pick a medium case or even an easy case but one that has a high ROI. Pick something that is very repeatable but that, if you can get a bot to do it, it saves you a lot of money. That would be your next use case. And that could even be your first one if you can't find something small and simple. Once you get the experience, you get it under your belt, move on to your more mature use cases.
The tool is flexible. The tool is very easy to pick up. I am concerned with some later cases though. When we get into some complex business logic or processes, I'm not sure how it's going to handle heavier business rules, so we'll have to wait till we get to that point and we'll have to hope that our customer relationship with Automation Anywhere will help us with more complex or tricky resources.
In our organization there are 16 of us using Automation Anywhere, and we'll be at 20 by the end of January. We'll see where that number goes. And that's only been the last six months. That's a big number to put on there, where we have a bunch of stuff going on and are trying to keep it in control and figure out our center of excellence and our standards and our practices. I know the other companies may go in and throw big numbers at it but we're at 20 and that number is only expected to keep growing. That's going to be limited by how many projects can be done, how much money there is for those projects, and how many people as resources we can find. Right now, our users are developers, testers, administrators, and we have a couple of project managers who have a limited administration view into it.
Maintenance depends on the bot itself. Initially, whenever we start with a bot, we always have one person dedicated to it. Depending on the what was found, one person for maintenance is usually fine and that usually dwindles. You'll typically end up with one person who has multiple bots or automations that they are maintaining, as time goes on. If anything, it's less than one person needed for maintenance.
Overall, I would rate it at eight out of ten. It's a very simple interface. It's a very straightforward approach. You can very quickly get in and get some proofs of concept going. It has logging and some reporting. Some of the things against it are when I compare it to where I came from before and some of the features that OpenSpan had. For a developer doing coding and debugging - developing the code for reusability and debugging the code - OpenSpan had some advantages that you can't easily overcome in Automation Anywhere. That would be the two points off in my rating of Automation Anywhere, not that I would give OpenSpan a ten either. I would probably give that an eight as well, for different reasons.