Automation Anywhere (AA) Review

Enables me to mimic human behavior on a screen, but takes a lot of memory and CPU resources

What is our primary use case?

My primary use cases are where I'm dealing with a lot of raw data extraction and transformation so that the data can be used by other systems.

An example would be getting the data out of PDF files, transforming it from semi-structured to structured and putting it into an extra-stable system like Excel or a CSV so that it can be used by other systems.

How has it helped my organization?

A lot of clients I work with have legacy systems and a lot of API access is not available. Some of the systems might be running off a server located somewhere else, while some would be running on a mainframe and I'm actually restricted to working with the screen. So these clients have a very good use case. If I'm working with the screens, Automation Anywhere really does help me because it gives me the control over the screens. If you are not looking at integrating legacy software, Automation Anywhere gets the job done. But if you need integration then you start looking for other RPA tools.

It definitely saves time and effort. Improving the workflow, that's not something Automation Anywhere provides. That's a different challenge altogether - to do a business process improvement so that automation gives you even more value. That type of process works in combination with Automation Anywhere, but it's not a part of Automation Anywhere. So the process improvement is separate. We optimize the process and then we run it through Automation Anywhere.

You can probably use any similar tools. But Automation Anywhere is one tool that actually gives me automation capability right at the start, without worrying about process improvement in the first place. I can just act like how another person would. Whereas if I do a little bit of process optimization, I can use another tool also. If I get access to APIs, I might use Blue Prism. If I get access to web elements, I go the way of UiPath. If it is a human-mimicking behavior, that's where I use Automation Anywhere.

What is most valuable?

What I like about Automation Anywhere is the object cloning and the way you can move the mouse and either go to a particular point or go to an element. That's very easy and intuitive in Automation Anywhere.

It gets the job done in terms of getting the data out of the pages. Although I have other tools, I still have this habit of going through the clicks. If you're going through the clicks, Automation Anywhere is the best.

Let's say you are on a website. You move the mouse around. You click on certain places. Automation Anywhere is better at that because you can adjust the screen directly or you can adjust the element. Whereas in, let's say, UiPath, it's a little bit complicated on the inside because there isn't a direct command for that. I have to go to a web scraper. In Automation Anywhere, I have the direct command to move my mouse. If I need to mimic a human behavior, I will use Automation Anywhere.

What needs improvement?

Automation Anywhere is troublesome for some people because of the way it is organized. It's organized as an encrypted script, which gets run via a domain-specific language which the user sees. 

There's a loss of overhead on the computing resources in Automation Anywhere. If you have an encrypted bot, the Automation Anywhere software has to read it first, decrypt it, and run it. So there is a potential that, if the logic of the bot isn't good enough, a lot of CPU and memory overload will happen. This is something that Automation Anywhere should look at because it takes a lot of computing resources. I have seen CPUs running at 100 percent.

In terms of additional features, if I am dealing with a dynamic workflow where the workflow might change based on the input parameters, then Automation Anywhere doesn't help me because the code is pretty much fixed. When I need those types of workflows I go to UiPath.

For how long have I used the solution?

One to three years.

What do I think about the stability of the solution?

It's highly stable. I'm pretty happy with Automation Anywhere. I'm pretty happy with the security of the bot. Once you make a bot, if you don't have access to Automation Anywhere, you really can't mess with the bot. I'm pretty happy with the stability. 

The only problem I have is that it takes a lot of memory and CPU usage for Automation Anywhere to do its internal encrypting and decrypting.

What do I think about the scalability of the solution?

I'm not yet happy with the scalability of Automation Anywhere. Scalability is good up to about 100 bots. Beyond that, I need to spread it into multiple sites, which means there is additional licensing cost.

How is customer service and technical support?

I would give Automation Anywhere's technical support about three-and-a-half out of five. They do have a lot of information published, but the response times aren't great within India, where we are located. I can't say anything about support in other markets.

One thing they need to improve on is the way they have been putting out so many terminologies in the market: IQ Bots, MetaBots. They need to define them properly, in simple terms. If I go to my client and say "IQ Bot" or "MetaBot," they don't understand anything. It falls back on us to figure out whether these types of things will be useful for our process or not.

Regarding their support, when they bring in these features, like IQ Bot and MetaBot, there isn't a lot of documentation that comes with them, which can cause confusion in the client's mind as well as the developer's mind. Even Automation Anywhere's guys aren't really clear on IQ Bots, MetaBots, and things within Automation Anywhere because, when we ask, they just give us the definition. That's not very helpful.

How was the initial setup?

It's pretty straightforward in terms of setting it up. It's not a lot of work, as compared to what you would do in Blue Prism, or even in or WorkFusion. I would say UiPath is the easiest to install and configure, while Automation Anywhere would be number two. Blue Prism would be way down because it's difficult installing and configuring it.

It doesn't take much time to deploy Automation Anywhere. We have built a script. We just run the script and within three or four minutes we are done. We don't really install Automation Anywhere by running it and then monitoring it, rather the script automatically installs it. That script lightens our load; we automate our own jobs as well.

In terms of implementation strategy, we have a set of requirements for the client's environment and hardware. For the environment, we need to look at the .NET framework, which version, the directory structure, folder structure, paths. And there are multiple items to be checked out regarding the hardware: We need to look at the RAM, the hard disk space, the connectivity. There's a lot of checking which must be done, but we do that through the script itself.

We have all the environments set up in one local place and once the script runs it goes and installs all the required software components. The .NET framework will be installed, the run-time engine will be installed, Automation Anywhere will be installed, and the policies will be set automatically for at least the end user, so that we can go and create more users.

Once we have the hardware, and once we are ready to install the environment, it takes us about 15 to 20 minutes.

For deployment of Automation Anywhere, we don't need a lot of staff. But when we are deploying the bots, we generally have an experienced guy who will look at the deployment of the bots within the Control Room. That's a different scenario altogether.

We don't require a lot of people for maintenance. What we do is, we transfer some of the load to the client's staff, in terms of monitoring and scheduling. Of course, we have one person keeping an eye on the entire thing. We have one person on a chargeable basis per client location. And this person also doesn't have a lot of work, so sometimes this person moves among the sites if there is no problem at all with the installation.

What was our ROI?

Companies now are not willing to put a large investment up front into these tools, unless the service provider that is developing the bots can assure that the bot will be successful and there will be certain savings. Clients are actually talking to the service providers first, rather than the RPA Software Vendors. It used to be that Automation Anywhere would go to the customer, convince them and sell them five licenses, and then the client would go out and start hunting for Automation Anywhere service providers or resources. The whole model has now changed 180 degrees. Now the clients are more interested in talking to consultants and trying to figure out which tool would be good, how many licenses they would need, what the scalability roadmap is, what will they be doing again in six months, 12 months, two years, etc.

It's hard to get a clear picture of the financial value that it can bring. For example, when we go in, we look at a process and we look at the value that automating the process can bring in, but there are other aspects which we look at, which are a kind of "chain effect." If I automate this, what else will break in the whole chain of processes? When there are processes A, B, and C, if I automate B, either A or C or both will feel the heat from this automation effort. If A and C are not conducive to screen-based automation, then I am in a fix because I can automate B using Automation Anywhere, but for A and C, I might need to use something else.

That type of analysis is now coming into the picture. Earlier, it was: Pick a process, automate it, feel the benefit, and then go for another. That is one reason why now we can also recommend hybrid models where multiple tools could be used via a single interface. We have to build the interface to Automation Anywhere and UiPath, or Automation Anywhere and Blue Prism to get the job done. That becomes an additional cost to the client.

What's my experience with pricing, setup cost, and licensing?

If you look at the capital expenditure, Automation Anywhere is number two to UiPath. But if you take a long-term view, on a scalable model of a large number of bots coming out, it slowly goes on to become the costliest tool. There is something they can do about that.

I did a cost comparison on short-term basis, long-term basis, CapEx versus OpEx, and Automation Anywhere is the costliest. Surprisingly, Blue Prism becomes the cheapest, if you look at the long-term view.

That's because of the licensing terms, the pricing policy, and the engagement models. Blue Prism doesn't want you to buy just one license. They want you to sign up for the long-term, for at least a minimum block of ten licenses. Automation Anywhere can give you a single license, so the capital expenditure is low. But as you go on, the OpEx, the regular increase in the number of licenses and the price per, starts to add up.

The capital expenditure goes out right at the point of buying the tool. For Automation Anywhere, I would need to spend $20,000. UiPath can give me something for $6,000, while Blue Prism will come in at $300,000. If I'm just experimenting, or I don't have a need for a large number of bots, or I can optimize my design to run bots sequentially on the same machine, Automation Anywhere vs UiPath is quite comparable.

Which other solutions did I evaluate?

Every tool has its own drawbacks. Blue Prism would probably be an eight out of ten, but the nearest comparable tool to Automation Anywhere would be Softomotive WinAutomation. They both work on the same principles, although the internal storage is different. WinAutomation also works on a domain-specific language system, and I would rate it a little notch higher than Automation Anywhere and give it a seven-and-a-half out of ten, but they are all in the same category. I don't really see any of them getting a ten, on my scale, right now.

UiPath can do wonders, but the technology is old. If I want to do machine-learning, I can't do it with UiPath. I would have to create another "open UiPath" for myself to be able to use machine-learning and artificial intelligence libraries which are there in the market, because I can't use them with UiPath. That's where UiPath also loses a couple of points.

What other advice do I have?

It's a very dynamic market and everyday new tricks are being discovered. My advice would be: Look at your process. If your process is screen-based, doesn't have a lot of things to do with APIs, go for Automation Anywhere. If somebody's looking specifically to implement Automation Anywhere, irrespective of what process they're automating, I would probably call it a bad move.

Role-wise, we follow our own system. We have a solution designer and we have an architect. These two guys work hand-in-hand, from solution design to a technical architecture of the Automation Anywhere bot. Then we have developers who develop the system. And we have the leads, of course, who are managers. They are senior staff who understand how the bot code is to be published and released into the Control Room. Most of the time, it's the solution designer and the architects who are critical for us, rather than the developers. The development part is easier than the design part. Designing automation takes a lot out of us.

In our organization we have 42 people, and most of these are multi-skilled on multiple tools. We do only specialized stuff, so some 20 of them would have been working on Automation Anywhere at some point. We use multiple tools. We are tool agnostic. We figure out which tool to use and go with that tool.

We don't have plans to push future usage of Automation Anywhere, most importantly because of extensibility of the tool: I can't extend it. So we created a workflow tool for ourselves similar to UiPath, but it's open to extensions. I don't see a lot of projects happening on Automation Anywhere for us unless the customer asks for it. In the Asia-Pacific market, it's either UiPath or Automation Anywhere. If they don't have a tool then, of course, we'll have to look at the type of project and recommend a tool.

I would rate Automation Anywhere at seven out of ten. The architecture is great. It's only the way they have tried to protect their own bots that is causing them to cannibalize themselves. Otherwise, it is great software. It works on a domain-specific language. You really don't need to understand .NET or Visual Basic or C# to work with it. The domain-specific language is more like English. They have done a great job making something, but there is a big scope for improvement if they want to really unsettle the other guys.

In my opinion, instead of sitting in their offices and not conversing with people out there, there are a lot of things Automation Anywhere can do if it listens to the people who are actually evaluating it, using it, and are happy or unhappy with it. I don't really see a mechanism where Automation Anywhere can be seen listening to this feedback. Secondly, they should be more open about their roadmap and where they are going with Automation Anywhere. What I want them to do is to make some more noise about their plans, rather than their current situation, because customers are not looking to buy Automation Anywhere for the next three years. They're looking to buy it so that if their processes change or if Automation Anywhere changes, it can still be usable for their organizations.

I can't keep on changing tools. Let's say I use Automation Anywhere where it's obvious and then it becomes unsuitable, so I have to change to another tool. That rarely happens because the users are familiar with it and change is the biggest barrier. People don't want to change. And the cost of training is actually more than the cost of the Automation Anywhere tool itself. You need to train different people with different skills, not only in Automation Anywhere but for every tool. You need different skills and different people to actually make the whole thing work.

**Disclosure: I am a real user, and this review is based on my own experience and opinions.
Add a Comment