What needs improvement with Kryon RPA?

Extracting meaningful data requires significant effort. You need a dedicated business analyst who can understand the output.

The process discovery tool of Kryon RPA could be improved because of its limitations. I'm looking forward to the new release, which will fix these limitations.

Areas, where I see some room for improvement would be the use of reading details from an Excel file. Most of the time, it's difficult when I say looping through a huge data set, and it's very difficult for the bot to interact with those data sets unless it starts offering the Excel file. As of today, from the reports I've worked with, I noticed that most reports are empty columns and two rows. So if Kryon can associate with that and extract those deposits into a variable, and the tools can easily fit out to column dash and two rows and columns, it will be nice. But as of today, the bot doesn't have the functionality to detect variables. We have to do it on the Excel file itself. I also noticed that when it comes to the use of bots, not everyone has the same knowledge, and it can be difficult to do certain research unless you have advanced knowledge of what you need to do. I also noticed that in the community, most people don't use it, and it's quite difficult when you need certain support.

What needs improvement in Kryon RPA is the support. Sometimes, the solution has problems, so I contacted support, but I didn't receive immediate help from the team. You have to wait a while before the problem gets solved. This happens a few times in terms of getting cases solved. It would also be better if Kryon RPA had a cloud version because currently, from what I know, it only has an on-premise version, though I could be wrong about that. There isn't any additional feature I'd like added to Kryon RPA because it's a good service, but the support is very, very bad, so that area could be improved.

Kryon RPA's integration with other solutions could be more seamless.

It could be more reliable in spying HTML. I did expect this to be better in the 19.4 version.

The product requires some more time in development. It still has some bugs and some things to work out. They're constantly releasing. They need to continue working out the general issues of using the product at scale, first and foremost. I've talked with some of their product teams about features that I would like to see. With the Process Discovery, they're collecting a lot of data and they're spending a lot of time in the algorithms of the data. But one of the things we've talked about very recently is being able to aggregate or summarize what a user is doing in a day. That summary wouldn't necessarily be in the context of a process, but in terms of the applications they're using, the timeline of what they're doing, the number of clicks per hour. That sort of information could be very valuable.

At first, we had problems with the object detection but in this latest release it has been better, and I think it keeps getting better. They also came and trained us a little bit so that helped. There are some known defects that are on our list for the next release, but that happens. It's technology. They're aware of them and they're working on them.

Process Discovery currently is not working for us. It's not giving correct information, correct stats. It's an issue that is currently open with Kryon as a ticket that they are investigating. In our instance, it did not function as advertised. As a result, business analyst hours haven't factored in because Process Discovery is generally the piece that's supposed to save business analyst time. As of right now, we've saved no time in that area. Also, troubleshooting and debugging could be improved. It's missing a proper debugger. It's very difficult, based on the log files and the fact that it doesn't have a proper debugger, to troubleshoot issues. It will sometimes end with no indication as to why it ended and it becomes a bit of trial and error. Obviously, that is not a great use of time. If there were better debugging tools and logging, we'd be able to find the issues more quickly and fix them more quickly.

This need for improvement deals with the Office APIs in it. There have been some times when dealing with the macro interactions that we have had to build our own way of executing those macros rather than using the inbuilt tools provided by Kryon to execute them. I would like to see some improvement there. We have also run into instances with some of the Office integration commands that are available, where we have had occasional troubles. They read the data once, hold a file open, or do something in the background. It would be nice to see the handling of this improved. This should probably be relatively minor improvements. I would like the pool of advanced commands to increase. It does have a very solid base with a wide range of commands ranging from date manipulations through to SAP integrations. It needs to shore up those commands, harden them up a little bit, and maybe expand upon them. This would be something that would be nice.

While it does help reduce manpower, we still do require manpower because there are some processes and steps that we cannot do with automation. It may be too sensitive to be done by automated processes. In those cases, we still need manpower. It does help to identify which tasks can be automated based on Kryon functionality, until you have close to 50 multiple tasks from one client. As soon as we automate any one of our top volume drivers, we try to understand how we can automate the next one orthe next highest volume driver. We have tried to use Kryon's functionality to the max, but still there are some limitations that don't allow us to automate all the tasks. E.g., anything that involves free form text is a major drawback. We use forms which get their data from our customers along with their account details. So, their form details are always the same. They don't change the fields, only the values are different. If it were an email, every customer would use their own words to describe the problem. It's difficult to write simple logic that is common for all types of words. I think free form text requires artificial intelligence. However, the lack of AI is not a big issues. Another major drawback is OCR. We are not able to read scanned documents correctly in a reliable way. There is always some margin of error. Some of the processes require us to read scanned documents, and you need to ensure that it is 100 percent accurate. Without that level of assurance, you can't automate such tasks. Our client uses Citrix NetScaler Unified Gateway. It is a virtual machine. All the tools that we use run from the client's location and through a virtual machine called Citrix. It is projecting the remote screen onto our screen. Kryon is not able to identify individual applications behind Citrix. All it makes is a screen by taking the image. So, Kryon's functionality is limited to screen reading. Because of this, we are not able to take advantage of the Windows functionality or web server functionality. like browsers. They are not able to identify whether it's a browser, etc. It can just read from the screen. Being able to identify applications inside Citrix would be a huge advantage for our processes. Their logging features are minimal. Reporting-wise, there are some reporting options, but I don't think they are very practical from the point of developers.

Process Discovery is brand new. We are on 19.1 for it. I know on 19.2 that they change the architecture completely. From talking to some other companies that use this, it sounds like we are missing some pretty big features that we will need. We will be doing an upgrade here in the near term. In general, Kryon's upgrade process is basically uninstall/reinstall at this point. They don't have an easy way to upgrade the software in place, which would be an added benefit. The process is not difficult. We have just a handful of robot machines with high availability enabled, along with a couple of app servers and a couple of database servers. Still, that's 12 machines which all need to be upgraded, and that is no simple effort when you're talking about a full reinstalled software. If we fully utilized our licensing by scaling out, we'd have more than 30 machines. We would have to upgrade on the field. Then, it's a pretty big task if they release new features and things that we want to take advantage of going forward. I see an opportunity for improvement from them here. We are using it to do some automated reporting, and right now, we can't put images into the HTML formatted body of an email. We can either attach an image or embed a link to an image, but we can't just drop in an image into the HTML. That is feedback that I have given them. It would be nice to be able to have a bot take an image and paste it in as you would in Outlook. Otherwise, we have to provide either those images on a public website. If we want them just to be attached, it just doesn't look as clean. Especially people in our sales force, who are on their mobiles a lot, are not on our network a lot. These are a challenge for them when they just want to be able to glance at the report and go on with their day. It seems like a small problem, but it's limiting for us in some of the areas where we could deploy more of this solution. We have a feature request in for this, and I'm hoping it will be included in the future. How it delineates file names on email attachments could be better. It is a hard coded comma, and if somebody includes a comment in a file name that messes it up. It is such a ridiculous thing. Who puts commas in file names? But, you would be surprised, and it happens. This is another simple thing they could easily tweak. Their Tier 1 support is pretty basic. You either have to jump through the same hoops every time or escalate to a different Tier through your rep.

The read from screen functionality needs improvement We still do a lot of stuff in the user interface. Finding UI information and split UI need improvement. I would like them to add more functionality from Excel, especially when you pull data from Excel. We've had a lot of good success using this solution for the full-cycle of automation from the discovery of processes to turning on the automation and scaling it up. We would like to see having a little more specific documentation, or some of the examples be easier to find. One of the big things that we get directed to now is the Kryon Community page, but a lot of that wasn't in existence or publicized when we started doing our training and education. Now, we're stumbling through using it. For the majority of the time, we have either had to try and interpret what the instructions meant or do a service desk ticket. I have had two people go through the online training. My feedback would be that they need to focus a bit more energy on that. I would like to see that online training be a bit more comprehensive in the projects that they create, so we can say, "Here's what we have," or, "Here's what we need." I don't know if maybe they should sit down and try to develop "training" documents as well. Or, like dummy wizards, where you have to copy and paste this wizard, start from scratch and learn these steps, then you put it together to make it work, which would be similar to what we did with the in-house training. They could add a bit more to the online training too, since it's still relatively new for us. We don't have a "dedicated" team of people who do this. This has been a learning experience, because we only have a limited number of people who can dedicate hours to it.

What we are lacking is OCR, the ability to read text. We would like it to have the ability to take a page which is free text and analyze it, and then for the robot to know what to do next. This ability does not exist in Kryon products, so our company is now trying another tool which has this specific feature - an IBM tool which combines RPA and OCR together. Also, when the robot stops working for any reason - it doesn't matter if it's something in our systems or our infrastructure - it does not know to go back to the same point it was at when it stopped. We have to start from the beginning or to delete the report. It cannot continue from the same point. We would like the robot to know how to continue.

The environment of storing the variables could be defined as local, which will only be accessible within the wizard itself and cannot be read by the embedding wizard. It will be good if Kryon has a function to find/replace the keywords and collapse/expand the groups.

We use Google Drive a lot. If somebody sends us a spreadsheet or a document, it's going to be in Google’s format. It's not going to be in Microsoft Office format. But currently, there's no integration with Google’s documents. That would be a huge plus to have. The same goes for, using Chrome natively. There's an extension but it doesn't necessarily always work. We have to close Chrome and reopen it to get it to pick things up sometimes, depending on the page and the IFrames and things like that.

From our experience, the product is not suitable for end users.

With respect to web-based applications that we have, Kryon may not be able to do certain things in terms of technology and support. Also, there are limitations on integrations with other platforms like ServiceNow. There are some issues integrating. It's not a really an open system. The product, its design, etc., is really good, but they have to look at the openness, how to integrate with other products which are available in the market, as well as with our own solutions. You can do integration but it is not so easy. You have to build certain things to integrate. It's not like an open API is available. It's there but it's not really open compared to competitors.

The cognitive part, the OCR could use improvement, but I know it's on their roadmap. We need stronger governance to know, when a transaction is done, whether it worked or failed; strong logging. That would be helpful.

We have found some difficulties regarding the integration with legacy systems where the screen has elements in the Hebrew language. The connection to Hebrew, right-to-left systems, is more challenging. We are discussing development of the product with Kryon. There are some other gaps, but everything is in process for closing the gaps. The main issue, from our point of view, is the Hebrew support.

I would like to see a better OCR solution. That's the main thing: identifying unstructured data, like all paper documents and data from pictures. Also, because we are in Israel - Hebrew is a hard language - we'd like it to better handle Hebrew.

I think our version is two versions behind. I just had a talk about it last week with our account manager at Kryon and we planned together to upgrade the environment once per year. In my opinion, today, if possible, it would be better to have it done automatically, like an application on your smartphone. Or even if done manually, the upgrade should just be "next, next, next," and it's upgraded, rather than making it a project to upgrade. In the digital era, that is one of the expectations, that it would be easier. Besides that, it's all about functionality: What you can do and what you cannot do. I'm not very familiar with the competitors' solutions, so I'm living in the world or Kryon functionality. Sometimes I'm looking to automate something and it's not there. That doesn't mean that it's not possible. In terms of additional features, the sky is the limit. For example, we are working with Oracle ERP. It would be great if there were built-in functions or commands to integrate with the Oracle application, and not everything through capturing screens and going from there. Currently, there is only an integration with Office, mainly to Excel. I understand that Excel is much more popular than Oracle in terms of use and in terms of integration. For an organization that works with the Oracle application, I would love to have even a plugin or a tool kit for that - maybe to SAP as well.

I would improve the object recognition features, making it not linked to the screen resolution or the position.