Building IQ bots is time-consuming. Initially, script-wise, it would take a lot of time and also the testing of testing part of it. Probably there might be something that has been advanced now, but I'm talking about before six months.

Maybe there has been an enhancement, but during the time that we were working, the time that would take up to build the requirements initially would need a lot of time.

In future releases, Automation Anywhere should promote this product more to make it more visible. It should be promoted to increase its visibility and awareness.

Whenever there is a variation in the same set of formats, the product considers it an additional group of formats. It causes some challenges while predicting the output. We faced this challenge across multiple implementations. However, it occurs less often. It mostly does not affect us.

We face a few data extraction challenges when we use a huge data set with different formats. Some things do not get extracted. For example, when we receive a statement from the brokers regarding the trading details, the document might contain multiple tables containing data. The product is unable to extract such data.

IQ Bot could improve the documentation.

It's a very heavy product. The solution requires 32GB RAM.

It would be good if the solution could extract or capture details from unstructured data. It would also be good if the solution could incorporate AI capabilities to extract details from documents like annual reports or policy documents.

There's still room for improvement in IQ Bot because sometimes, my team has certain types of PDFs that need to be grouped, for example, five types of videos need to be sorted to get data, but it was impossible to do in the solution.

There's even an instance when I'm extracting data, for example, the invoice numbers, and I trained IQ Bot to do the extraction just for the invoice numbers, but it failed to do it. There could be some exceptions that the product developer is working on, but it could take time because sometimes IQ Bot works, and sometimes it doesn't, so the error isn't because of the developer. The error lies with the system.

Another area for improvement in the solution is that during installation, it makes your machine drastically slow. It requires you to have RAM of 16 GB and a 500 GB hard disk, but that's still not enough.

Slow downloading is also an area for improvement in IQ Bot because whenever you try to execute bots around 20MB to 100MB, for example, that will be downloaded from the cloud to local machines, and this process could take between seven to eight minutes, and this could be frustrating especially when you're trying to showcase it to a customer. Bot downloading into a local machine shouldn't be taking more than five minutes because it's embarrassing in front of clients.

It's good that Automation Anywhere provides a cloud version for IQ Bot, but if you're unable to execute it fast, then that's not very good. I was working for one client in Brazil and there was a time I was showing the bot to them and it was taking around seven minutes to get downloaded on the local machine, and I even tried to connect with the support team, but that still didn't get me anywhere.

If you're working on some code and you're executing a bot, you need to change one line and the bot will download completely, so it should be pipeline or SVN or Git. If you've downloaded a 50MB bot, you just change one variable name, but then you have to download the 50MB code again, and it makes you wonder why you need to download the complete code. You have to double-check if you already have the code in your folder, and why would you need to download it again. Only one line should be changed for the download time to decrease drastically, but right now, you need to be downloading the code completely, so if you're executing a 100MB bot fifty times, then that means it will be downloaded fifty times as well.

IQ Bot is not a straightforward solution, so my team is struggling with it. Currently, my company has been doing some R&D about the solution, and it has several main problems that need to be addressed.

What I'd like to see in the next version of the solution is support for unstructured PDFs because currently, it's a very good solution for structured PDFs.

The percentage of accuracy is very low. When I put documents in IQ Bot and the document format gets changed, IQ Bot puts that in the validation queue. If you are uploading 100 documents, the IQ Bot accepts only 60 documents. 40% or 30% of documents go to the validation queue. Accuracy is way low. They need to improve their accuracy level. It should be at least 90%.

Another major limitation is with the handwritten applications. If I write a new application with a pen and scan it and store it, IQ Bot cannot access that information. It doesn't accept handwritten applications. It should accept the handwritten application.

Sometimes, the quality is not good, but it also depends on the quality of the document. If the quality is less than 600 DPI or 300 DPI, IQ Bot will not process the document.

There are two areas where improvement is needed. Firstly, I find it difficult to select the key and value mapping area when creating learning instances. I suggest that the selection area could be improved. Secondly, IQ Bot should add more predefined headers for each type of document, and allow for custom headers to be added.

In the next release, I would like to add more features to IQ Bot, such as text classification, image classification, sentiment analysis, and natural language processing tools. I believe that these additional features would make things easier and enhance the capabilities of the tool beyond document analysis.

When you update to a new version, you have to migrate your applications. That should be automated because it should not affect your ability to complete the task. If the version of the application changes, then you need to automate some part of that again. 

The problem with IQ Bot is the price. With so many options on the internet, some customers, medium-sized companies, have no budget to pay for IQ Bot, and then they just migrate the IntelligentOCR and all the machine learning to data extraction to cloud providers, such as Foundation AI or ABBYY.

In terms of Automation Anywhere, I did not try it yet, however, until the IQ Bot 11, what I can ask for Automation Anywhere to improve is just NLP and NLU, to extract body emails or an unstructured text.

I don't know if we can use IQ Bot with another RPA vendor by just integrating IQ Bot through API. I don't know if it's possible. I'm working with UiPath as well and I don't know if we can use IQ Bot with UiPath. 

We'd like just a bit more improved technology. Sometimes we send documents with less than 300 DPI. I know that Automation Anywhere just said that IQ Bot 11 and higher versions are ready to extract the documents with less than 300 DPI, however, this is not true. I never could work with documents with less than 300 DPI.

The success rate must be improved. It is mostly used for printed documents. It would be helpful if it could also convert handwritten documents. The product must also provide AI features.

We find errors with this solution if we have a document in paper format and use a scanner to convert it to a digital document like a PNG. As an additional feature, the technology needs to recognize signatures. Currently, there are issues with the technology identifying a signature.

We also used a previous version of IQ Bot about three years ago when we started our pilot and experienced some problems. We had issues exporting an instance and could not import it into another environment.

We've been happy with the product as a whole. There are no missing features.

From what we are operating from now, we don't see any problem areas or anything that needs improvement. 

I would like love to see pre-trained models based on industries. If I'm looking at an insurance domain, if the UiPath, IQ Bot has an algorithm that is pre-trained with the various formats that are available in the market, that would reduce a good amount of development time.

The product is expensive. 

They're always trying to make it easier. The whole automation industry is moving into what they call "intelligent automation", where they have started to bring AI into it also. That way, based on decisions, automation will happen.

The process of manually correcting the data in the validation queue that IQ Bot failed to extract should be automated. IQ Bot's AI technology has to be improved.

While there is adaptability in terms of the data series, an issue that should be addressed concerns different sizes sitting on deck, for something that can be done respecting the terms of the sizing of the template that the bot should be. 

Another issue is that the user should be able to update the data libraries. When it comes to small matters, a person should not need to be at the mercy of an implementation partner or Automation Anywhere for the receipt of instructions to reach out to the technical helpers

For other matters, we worked proactively with IQ Bot to address pain areas that we had encountered in our prior experience, such as that of the design evaluation or of the work we had done with our hardware providers in terms of the quality of the scanning and compression technology. This we addressed because of the inordinate burden on hardware capacity when doing high image scans, something which tremendously raises the size boost. So, we have, over time, optimized that along with the Automation Anyway team.

For anything in a standard location, we have fixed the sizes and do examples in the time of the ULT. Our extensive proactive work with IQ Bot reduced our pain areas.

Furthermore, in the past, I had forgone solutions that are in the scanning mode, such as vendor invoice management or OpenText from this or from Kofax. We contemplated the advantages and disadvantages of a scanning solution, a matter we have been instrumental in addressing. We now refrain from trying to do every conceivable task and to scan the entire document.

It should have more artificial intelligence to capture information from a document. Currently, our clients are preparing invoices manually. They are not digital, so we need to scan all the documents, but we are facing issues in capturing all the information through IQ Bot. Therefore, we have to use other tools to capture the information, and then we use IQ Bot to process all the information. They need to improve the artificial intelligence for document scanning.

IQ Bot doesn't accept or manage unstructured documents properly.

Lacks sufficient flexibility in identifying a wider variety of documents such as those that have different dimensions and quality. It would be helpful if even low-quality documents could be identified. They also need to work on the deployment of backups.

There are certain things that OCR engines are not capable of handling currently. For example, OCR engines pick up a slash as a question mark. They also don't have the capability to handle subcolumns in a column. It would be really helpful if they can add more OCR engines with advanced capabilities to handle such scenarios. We need more cognitive abilities in the OCR engines to deal with the most complex scenarios. 

We also need more advanced capabilities so that IQ Bot can give better results in terms of PDF quality. This is because we cannot expect the same quality from all the vendors and customers, and there might be scenarios where we get some sort of scanned copies.

The solution did not allow the use of other OCR and their own was not accurate and could improve.

IQ Bot uses an API OCR engine. At times, they can't accommodate some text in Korean. Also, Korean characters are not cognitive at times.

It should be easier to install and configure. We had some issues during the initial setup, and we are resolving them with the help of their support. Its initial setup is complex. 

Their web pages are not really structured. Because they have multiple products that are more or less similar, it is sometimes hard to understand which documentation you're actually reading and using. Their documentation and web pages should be structured in a better way.

In terms of new features, it is a little too early to provide inputs. I need more real business cases to see what is lacking. We haven't yet got those. We are currently testing it from a technical point of view.

The way this solution creates invoices can be improved. It is slow. Sometimes, it is also very complicated because you need to apply some kind of filters and effects over the images, which makes it a little bit complicated. They can reduce the complexity to manage the fields or incorporate other features. They can provide an easy way for developers to customize and program fields.

The accuracy of the OCR engine can be improved. It would be great if they can also provide an ICR engine and an option to choose between OCR or ICR.

There should be built-in or native intelligent processing capabilities.

IQ Bot needs to scale up further because many other industry products that are capable of capturing handwritten information are doing much better. So, it would be better if IQ Bot spent more time on handwritten data extraction.

Some of the features that are currently existing could certainly be enhanced. The ability to define the coordinates on the screen, on the documents that are being scanned, was a little less when we did the proof of concept. These kinds of features could be enhanced a little more. It looks a little rigid at this point in time, but it could be improved.

The first place it could be improved is its price - it is too high. I'm having a lot of difficulties moving it to the customers. If they could fix that it would be awesome. Because the Automation Anywhere is hardware-intensive, it requires other set-ups, a control room, and it's already taking 16 GB of space. Then if I move to IQ Bot, and I have to set up another server for it, it becomes pretty expensive. And that is where the hardware cost comes in. So I think that if they could reduce the cost thing it would be really helpful.

I would say it is the best choice, especially if you're already on Automation Anywhere. They are working at implementing third-party tools. Right now, I am with the company, but before that, I was with another company as well as IQ Bot. A lot of people have purchased a lot of license work but they have not actually put it to good use, or they were unable to. So, it is a question of quality control. But then again, from a CTA perspective, I would say I know it is possible to increase the optical features, as well.

I am not referring to just character recognition, but for example, let's have the scanner import that image properly. Sometimes in a test it happens that the picture is a bit clearer. Or let's say a company logo. If a print is misaligned, it can figure it out. So I am more than happy regarding their automation technology.

The price of IQ Bot, packaged with Automation Anywhere, is too expensive for some companies and they will not use it. The price should be lowered.

Migrating from the previous version to the current version is a little tricky. We could not find proper documentation or guidelines to help us migrate the bots from the previous version, version 11.3, to the newer version A2019. 

As this kind of documentation is lacking, whoever wants to migrate from the previous version to the newer version will face some difficulties. That's one thing they can improve on as it'll be helpful for the users.

We are working with IQ Bot, but we are not impressed with it. It works well with static documents but not otherwise.

It was challenging to have to do the pre-processing and tax inputting on our own. They didn't have a feature through IQ Bot.

The way you implement the pre-processing needs improvement. In many cases, the invoices that are received have the fax cover page, and with terms and conditions page. We have to input all of the pre-processing. It would be better to have it as a part of this solution, otherwise, every single invoice would be treated as a new document, even if it comes with terms and conditions and a cover page.

They still have a long way to go when dealing with a structured and unstructured environment.

Features are good, but when dealing with pre-existing models such as purchase orders, and invoice,s for example, the capabilities of the solution need improvement. 

It's not necessarily the IQ bot but the Automation Anywhere as a whole, I find it slightly complicated to use as compared to a product like Uibot. I would say it's not intuitive from a user standpoint. This leads to a specific specialist to be part of the team to work on it. Because otherwise business processes are best understood by the business people and the IT applications are better understood by the IT people. But when it comes to these kinds of automation, it is the business people that should be able to do it on their own. For that, it has to be intuitive. So that is one key concern that I found. It's not that intuitive for a person to work on it. The learning curve is quite high. However, adding the layers of UI would not mean add in a lot of layers of abstraction because that also adds a problem. The flexibility drops. The machine learning techniques like we have for these bots are slightly limited as of now. Maybe they're already extending it or maybe in the latest version that I have not used will provide extensive machine learning algorithms that can be integrated with the bot. But that was my only space which I see that could be better. Very soon features would be more, on a large number or more machine learning algorithms and specifically there is what we are researching right now, ourselves, with is machine learning. But, there is a subset of it which is self-learning. Self-learning is going to be a key aspect because as we start automating the processes, the processes also evolve and sometimes we need to read deep parts instead of that, there should be the capability of self-learning algorithms available. However, it's easier said than done and I can understand it's very difficult. But that is the key piece which I would like to see do better right now.

Because there are certain restrictions or limitations with IQ Bot we started to explore ABBYY FlexiCapture.

In terms of extracting the information, we had a lot of issues with Textract OCR, and we discovered that it wasn't very good.

With ABBYY, we were able to achieve some solid outcomes; I won't say they were significant, but they were satisfactory.

The groups that are built based on templates and patterns, it's a tremendous undertaking, and we found it a little tough in terms of maintaining those groups and, in particular, training. In terms of document training, we found it to be a little difficult.

We are looking for better results when it comes to scalability.

Machine learning capabilities would be beneficial to have.

My biggest challenge is the monthly increase in volume. Because there are so many new organizations being formed, I find it difficult to train and maintain them. There are numerous duplicate files in the system. Cleaning those is becoming a difficult task for me. Training is becoming more difficult. 

They are not able to explain why distinct groups are formed when the patterns are the same.

We have plans to scale this across other departments. I have uploaded almost 14,000 purchase orders and I have almost 600 groups, with several duplications. Cleaning it up, and training the groups is a significant challenge. This is definitely an area that requires improvement. 

Another concern is the technical support, specifically, those responsible for the tickets, are not very helpful. I have to contact them, spend hours on the phone with them without getting my issue resolved.

It would be better if they had some in-built formats for nonstandard requirements. Currently, the invoice has a thing, purchase orders has a thing, and other details of employment and onboarding are all easy. But if I have to put something out of the box, let's say if I want to read a rental agreement or if I want to read a property agreement, there isn't an option for that. 

Initially, there was one issue when we were trying to switch from a development environment to a production environment. We needed to transpose or expose a bot to the production environment. There we had some issues where the in-built input feature stopped working for extended periods. It was an API to kind of input the versions from development to the production environment. But I think it has been fixed in the upcoming one.

I think Automation Anywhere has a partnership with Google. I hope they integrate Google's OCR engine in some way with IQ Bot because the OCR engine that Google produces provides an in-built accuracy of close to 99%. That is the integration we are looking for, but we need to pay for IQ Bot. If they can integrate both and come up with better pricing, then it will be a huge boost because, in that case, the quality of an image or PDF quality is taken completely out of difficulty as Google tends to extract the data. That is something I am looking for if this happens.

The solution could improve by having more domain-specific templates, such as for documents other than invoices or contracts. For example, having templates for IDs or Passports.

I think improvement could be made where handwritten documents are concerned. If you have a printed invoice, it's easily identifiable, but if someone is writing the same thing in their own handwriting, then identifying it is much more difficult and less accurate. I'd also like to see more support available for our customers. 

This solution could be improved by making it possible to add handwriting and signatures. 

The solution's predictive model can be improved in terms of accuracy. It takes a lot of training to get it right. It is hard to customize IQ Bot because each client has different types of documents and requirements. It takes time to get accuracy, particularly when working in a country like India, where many documents are scanned.

If they could improve the usability to train the models that process the farms, that would be helpful for our clients.

It would be helpful if the solution was better at reading handwriting. It needs better handwriting recognition.

In one of the applications we were working on, we found challenges. Automation Anywhere enterprise 11 has troubles interacting with a Pega product that we are using. Consequently, there were projects that we were not able to complete.

One of the challenges that we faced was that the Automation Anywhere log files were being created and not cleaned up, which would eventually kill the bot. We began a practice where we cleaned out the log files on a daily basis, which improved the bot.

Automation Anywhere does not have support for all legacy systems. In the insurance industry, they mostly work with legacy systems and will not move to future technologies very soon because of the security measures, cost constraints, and other reasons. In these cases, we do not see improvements in terms of efficiency.

I would like the intelligence to improve so that IQ Bot can handle vastly different formats of documents that we get from the customer, in terms of reading different fields. Having it even more intelligent would be very useful.

Better customization in terms of mapping would be an improvement.

API configurations through control room could be improved a bit. 

If Automation Anywhere can be provided at a lower cost, it would be more beneficial.

Technical support could be improved.

Also, IQ Bot could increase the handwritten document reading ability.

We're not so happy with the cloud version, which is more complex than the on-prem, so it's more difficult to utilize. IQ Bot's consistency with parameters and its OCR engine could also be improved.

The quality of the service needs to be improved.

Having a complete cloud version, as a SaaS, would be quite nice.

A community version of IQ Bot should be released so that more developers can gain experience with using it.

It would be helpful if the IQ Bot OCR could be integrated with third-party solutions like FlexiCapture.

I've used versions 11 and 2019 of this solution. The problem with version 11 is integration. When you are using a conventional bot and you try to use IQ Bot development, you don't have a specific feature for integrating these two pieces or components of Automation Anywhere. I was seeing some issues while trying to, for example, send some PDF to an IQ Bot processor and get a response for that process. So, there is a problem with integrating the conventional task automation and IQ Bot in version 11. In contrast, version 2019 already has the integration feature, and we can easily integrate two components. There are some features that must be there in version 11 but are currently missing.

There is no total control over some features in IQ Bot. Some variables cannot be parametrized when using IQ Bot directly. The AI features that IQ Bot has must be more open to being parameterized by the user. It would be good if Automation Anywhere exposed some parameters so that I can adjust them as a customer. This is an improvement that Automation Anywhere can do.

I would like to see some features related to deep learning in the next releases of IQ Bot. These will be some interesting features to see.

The log scripts can only be written in Python, there is no compatible language.

The cloud version is not good and the fees are expensive.

The Auto Logic creation is available as a plug-and-play model, and this should be available in IQ Bot.

The product algorithms could be improved to increase accuracy. We currently need to make regular adjustments, constantly retraining the product to cover all cases.