ABBYY Vantage Reviews

We are using the solution in our company for extracting details from invoice copies.

In our company, using the solution, we have seen that overall automation is quite easy. The output and time for developing the process are quite less.

It's worth noting that ABBYY Vantage employs both AI and machine learning techniques. Additionally, integrating with ABBYY Vantage is easy, with LinkedIn options available for use through the platform's built-in activities.

The customization can be done using ABBYY Vantage. But I don't have complete knowledge of its pricing.

Customization in the solution needs to be modified. In the future, I would like to see the product have better customization functionality.

I have been using the solution for a month. Also, I am using the solution's latest version. My company has a partnership with ABBYY Vantage.

Stability-wise, I can't rate the solution since it is still in the PoC phase. Once the products which we are developing go live, we can figure out the stability of the product. As of now, I rate the solution's stability an eight out of ten.

Scalability-wise, I rate the solution a seven to eight out of ten.

The technical support is helpful. I rate the technical support a seven out of ten.

Neutral

Previously, I used to use FlexiCapture. Still, we are using FlexiCapture in our company. ABBYY Vantage is user-friendly. Also, it has custom activities, and integrations are quite easy.

The deployment process can be done in a few minutes. So it can be done in less than 30 mins.

We were able to do the installation with the help of our in-house team. We already have a platform team who actually helped us with installation. So, I just started using the tool last month. I wasn't involved in the installation since I am a core developer who works on the design. Approximately 10-12 members are using the solution in its PoC phase. Later on, it may expand to thousands of users.

The cost is per page we use. So, the number of pages we use will be its cost.

I would definitely recommend the solution to those planning to use it. I am yet to explore the solution's load-balancing feature. I rate the overall product between seven to eight out of ten.

To be more concrete, I rate the overall product an eight out of ten.

Our company uses the solution for processing invoices that follow a particular template. We have some standard invoices on a monthly basis and use UiPath to download them from our suppliers. The downloads are extracted and exported to the solution.

We previously used the solution's on-premises version but it is no longer supported. We used the trial cloud version until we bought our license one month ago. 

We are in the development phase but once we go live, we will have many users for the verification station.

The solution has its own language that can be configured many ways and with higher accuracy than other products.

Programming interfaces should include intelligence or documentation. It is difficult to find documentation and implement without some information included in the backend scripting pages.

The support network is weak and needs improvement.

I have been using the solution for one year.

The solution is stable for documents that are standard format.

The solution is scalable.

The support is weak and it is difficult to find answers for day-to-day problems. There is no forum for answers and not many users are familiar with the solution to help with issues. The provider should have documentation for everything but it is not reaching end users.

Tickets are not responded to in a timely manner and it is difficult to get email responses or escalations. There is no standard format for support and they just respond whenever they like. In contrast, UiPath helped with anything needed.

I rate support a four out of ten.

Neutral

We used UiPath's Document Understanding but it had issues finding items that were marked in particular locations.

Accuracy is comparatively higher for the solution. You have the option to write rules in the backend which can be manually verified if a document does not satisfy conditions. Once verification is complete, the document feeds back and the solution picks up the rest of the processing. This flexibility was not available in UiPath.

UiPath also required too much user involvement at the time of our implementation, but I believe this has improved over time.

The cloud setup is straightforward and only requires configuring the development environment which includes flexi layout studio and product setup station. Any security concerns can be addressed through an SSO or AD integration.

The on-premises setup is a bit complex because it requires understanding connections.

We implemented the solution in-house.

Maintenance and updates for the cloud version are handled by ABBYY. We receive notification of updates in advance. Typically, an update requires some upgrade in the development environment to ensure there is no impact to users.

Our license is $6,000 and includes unlimited users for the verification station.

Licensing is a bit on the high side.

We looked at ABBYY's invoice processing but were not satisfied with the results so we moved to the solution.

The solution is the best for use cases with standard document templates. If your documents do not follow a particular format, then you cannot use the solution. 

The features for AI and processing lag behind so I rate AI a five out of ten.

Overall, for our purpose with standard document templates, I rate the solution a seven out of ten. 

We are a logistics company, and, due to the fact that a lot of documents get generated on a daily basis, documents need to be processed and data extracted. That's our primary use case for this solution. We use it for a lot of data extraction.

The fact that IBM Datacap uses ABBYY's core engine for OCR, is quite useful for us. It makes them one of the key players as far as OCR extraction is concerned. 

The fact that they've packaged it very well in the sense that you have these field extraction batches which enable you to auto claim the documents, and you have these templates which are structurally laid out, which enables you to mark a specific zone from which data can be extracted. 

It saves us a lot of manual work.

The automation on offer is good.

Technical support, for the most part, is helpful.

A lot of things could be improved, especially if you look at the number of support tickets that we have created. Looking at those would indicate the things that can be improved. 

For example, more often than not, we are forced to rely on template creation in order to bring the accuracy of extraction to 100%. 

Even though the training works, it's not as good as it is made out to be. In our case, we've got thousands of suppliers and it is just not humanly possible for us to keep creating templates for each one of these suppliers. What we would need is an out-of-the-box extraction to be even more accurate, or for the training to improve in such a way that just the training should be able to give us a reasonable amount of accuracy. I know that 100% is too high a limit to aim at, however, even if we get close to 85% to 90% margin, it would be much better. Something that could do that out of the box would solve most of the problems.

ABBYY introduced a reporting data warehouse last December, however, that reporting data warehouse has its own fair share of problems. Something like accuracy and completeness of out-of-the-box recognition is something that should be available as part of your default reports in the tool itself. Right now, that is missing. 

ABBYY provides a document definition for an invoice, however, that invoice is a supplier invoice. In our organization, we deal with commercial invoices most of the time. The problem there is that supplier invoices. If you put them into the system, it gives you around 80% to 85% accuracy out of the box. However, with commercial invoices, you only get around 60% accuracy out of the box.

Based on the input that we have provided, ABBYY is now working on including the commercial invoices as well as part of that default invoice document definition. They are working on that as well. There are a lot of minor things that have come up. For example, when you create a template, ABBYY claims that you can export the default template that is generated as part of the training, and then make changes to that and then import it back. What we found was that the concept applies only to certain things. The training of the line items cannot be exported as a template. They said it's to do with the way machine learning was being applied and all that. However, these are all things which kind of increase the development time.

I have a couple of years' experience with this solution. I've been working with it since about 2019.

The performance can improve. Even though we have tried out the solution's provider, ABBYY, in terms of having more processing stations, all still boils down to the fact of how your application is configured. For example, in our case, we have about five processing stations in production. When we started off, it took almost 60 seconds to process one page because of the complex nature of our application. We had around five document types, and each type has to be classified based on a set of keywords, and then you had to apply the document definitions on top of them. Then it goes through this concept of the generic layout and the additional layout and so forth. Almost every page used to take about 60 seconds on average. Then we made a lot of tweaks to the application and then we brought it down to around 22 seconds per page.

That said, even then, when you're talking about the space that we are in, the domain that we are working in, sometimes a 100-page document has to be processed within around 30 minutes. If you take a 100-page document, only for processing that itself, it takes almost close to an hour. Then, what happens is that additional stages like verification, et cetera, by the time the data is exported and uploaded into the target system, it's taken too long.

With the normal ABBYY support, you go to support.abby.com and you raise tickets. The support is almost immediate. There are constant responses that come to you. However, what I've also seen is that there are a lot of complex problems for which we have not gotten the solution we need. While the response has been great, if you asked me whether all the queries for the problems that we raised have been solved, the answer is no. On the other hand, we also have a chance to interact with the ABBYY professional services team directly due to the contract that has been established with our organization. They are a wonderful bunch of people and there's this personal rapport that we have with them. We get in touch with them on a weekly basis and our query is resolved.

In our organization, it is the IT team that takes care of the installation. I do not have much information about the process and I wasn't really a part of it. 

We're just a customer and an end-user.

I'd advise users that, when you start off with your first project on FlexiCapture, choose a project which has at least something which can accommodate a very high pattern. For example, if you're going to go for a use case wherein documents have to be processed within the next 30 minutes, you're going to face a lot of problems. ABBYY is a wonderful tool, however, it has got its own set of constraints. It's very important to understand the constraints of the tool.

The strength of the ABBYY FlexiCapture license is the OCR engine. However, if you're extracting it to process millions of documents within a very small amount of time, that is not going to happen. 

It's very, very important for the final operators who are going to use the tool to understand what OCR is all about. The problems that we face are not technical at all. It is about trying to convince people, people who have been doing operations manually to change their processes. If you have a situation where people are looking at documents, looking at the data, and entering it into a system, and then come and tell them that the solution will extract that information automatically and they just have to verify that information, they'll need to change their approach. 

It's very important to get the business team who will be doing the verification into a meeting, in order to help them understand what OCR is, and also its limitations. 

You will need to specify the stock field and the zones. OCR is able to learn on its own and understand how to extract data from it. However, there are certain things that you will have to teach people. You have to tell them the fact that a page is converted into black and white, and that the gray areas either cannot be black or white as certain characters that are being recognized may turn out to be something else. OCR tells you the level of confidence based on what gets extracted. All this has to be translated to the business team so that they understand the tool and its limitations. If you do this, you can ensure the success of the project.

Therefore, while ABBYY, as a tool, is great, there is a lot of work that needs to be done before you start implementing it.

I'd rate the solution overall at a nine out of ten. We have had to initiate a lot of fixes, however, overall, it's quite good.

Our primary use case of this solution is for document processing. Whenever I receive an email from a vendor or business it gets processed to the ABBYY server. I've created a layout in FlexiCapture that recognizes the template, extracts the field and puts it into the Excel format. I mainly work with purchase and invoice documents and ABBYY checks the validation conditions of an invoice. I'm the manager RPA developer and we are customers of ABBYY FlexiCapture.

The advantage of the solution is being able to identify and extract the element you need from a document. The template can be customized for this. We can create our own rules to validate some values before exporting the results. 

I would like to see consistency improved. If the same document is uploaded twice, I might get a different outcome the second time. Machine learning needs to be improved. We have some issues with scanned documents that may have a quality less than 300 dpi and the quality isn't high enough for ABBYY to accurately read it, and it therefore requires manual intervention. There is scope for improvement which could include AI modelling and training. If AI is there then it will identify it in a better way. There is a field training option that already has an ML algorithm for it, but it's not much use. Hand-written documents have less than 50% accuracy. 

I've been using this solution for a year and a half. 

We haven't had any issues with stability, I think it's better than other solutions. 

I don't know about scalability because I'm the only user in the company. We deployed in the AWS server and we have a team who control everything. We currently have eight projects related to OCR which are in the pipeline. I have already trained two other guys.

There is a forum for support but not many people are active. We have a license with them so we will usually receive a response within 48 hours. 

The initial setup was okay, the difficulties came later with the poor quality of documents we received. 

We pay an annual license fee, our management team that deals with that. 

I did a POC on all the different OCR tools in terms of the level of accuracy I get from each document. ABBYY and Amazon Textract did quite well, whereas Kofax did not provide as good a result. Textract reads the entire document data and provides raw data output in the string format. With ABBYY, I can add a field into the template and then extract it but in Textract adding a customized field is not possible.

Anyone using this needs to know they won't get 100% accuracy. Human intervention is required depending on the quality of the document. Clearly, if the quality is good, the results will be better. ABBYY has been on the market for over 20 years so they're well supported including all other RPA tools. Competitively, I think ABBYY is better than other OCR tools. 

It's good to test the solution with multiple samples. Once you create a layout, test it with multiple documents and see what you're able to achieve. If I see that 30 out of 100 documents are not processed 100%, I'll retrain it to improve the accuracy level for when the same document comes in next time.

I rate this solution a six out of 10. 

I mainly use ABBYY Vantage to process loan originations for cars.

ABBYY Vantage has reduced our data because it can classify and extract information from documents.

ABBYY Vantage's most valuable feature is auto-learning.

ABBYY Vantage's reporting could be improved - they don't have any out-of-the-box or detailed advanced reports.

I've been using ABBYY Vantage for about a year.

I would rate ABBYY Vantage's stability five out of ten.

I would rate ABBYY Vantage's scalability eight out of ten.

We've seen an ROI from ABBYY Vantage.

I'd give ABBYY Vantage's pricing a rating of seven out of ten.

I'd recommend ABBYY Vantage to other users and give it a rating of seven out of ten.

The most valuable feature of ABBYY FlexiCapture is the ability to create your own template on the fly. You can train and classify, you can work with those different templates without the need to do special programming for it or with minimum programming. Additionally, the solution is supporting the Arabic OCR, both of these combined features are the most important for us in the Egyptian market.

ABBYY FlexiCapture could improve the performance. We are looking for better performance because it can be a lengthy capture process and the accuracy needs to be enhanced when it comes to Arabic.

In the next upgrade, ABBYY FlexiCapture could add invoices for the Arabic or for the Middle East because it's based on European or American standards mainly.

I have been using ABBYY FlexiCapture for approximately four years.

ABBYY FlexiCapture is stable, we did not find any problems.

In our usage, ABBYY FlexiCapture has been scalable enough. We had a cluster-based operation.

I rate the support form ABBYY FlexiCapture a four out of five.

The support could be more responsive.

I have used OpenText previously and overall the solution might work better, but for our use case, ABBYY FlexiCapture has all the features we need that are important, such as Arabic OCR, and is priced well.

The initial setup of ABBYY FlexiCapture is easy.

The price of ABBYY FlexiCapture is better than some competitors, such as OpenText.

I recommend this solution to others.

I rate ABBYY FlexiCapture an eight out of ten.

We use the solution to extract the contents from documents like invoices and passports.

Verification Station is valuable. If the confidence level is lower than what we set, we can view it in Verification Station and correct it if there are any issues. For example, if the confidence value is 90%, and the confidence level of the extracted content is below 90, the documents will be pushed to Verification Station. We can check whether the document has been extracted correctly. If the values do not match, we can correct them.

It would be good if the product provided an AI-based prebuilt model that predicts and gives output. We can then verify and make changes if necessary.

I have been using the solution for almost three years.

I rate the solution’s stability a seven out of ten.

I rate the tool’s scalability a nine out of ten. Five to ten of our clients are using the solution. The number of users has reduced. There are other vendors who offer OCR solutions. Our clients are enterprise-level businesses.

Whenever we have issues, we raise a ticket, and we get a response from the support team.

Positive

The initial setup is not easy, but it is not complicated. It was pretty straightforward. I rate the ease of setup a five or six out of ten. The vendor provided all the documentation. We could raise a ticket if we had any doubts, and the team would help us.

Compared to most solutions, it is a good product. It has a lot of features. The extraction it provides is much better than any other vendor. Overall, I rate the tool a nine out of ten.

We use this solution to extract data, and to combine it with other solutions to provide end to end processes.

We like the fact that this solution is enterprise-ready, so it can be deployed on-premises, scale it, etc.

We would like the workflow building within this solution to be more transparent, which would allow us to model the process.

Also, the large number of components within this tool can make deploying and managing the solution quite complex.

Lastly, we feel that the usability of the interface could be improved by the use of cognitive capabilities, to make it more user-friendly. At present it is quite complicated to understand how to use this solution, and can take time to master effectively.

We have been working with this solution for two years, but we have been more actively engaging with it for the past year.

This is a fairly stable solution. In our experience there have ben some issues, but these have not stopped production. Instead it has cost us time in order to contact support, understand the issue, and then to mitigate future risk.

This is a scalable solution, as far as we are aware.

The support for this solution tend to respond in a timely fashion. However, more complex issues may mean a delay in replies.

Positive

This solution is fairly complex, even for expert users, to setup and deploy. There are a lot of components involved in the initial setup, and a lot of documentation to go through in order to deploy the tool successfully.

The licenses for this solution are payable on a yearly basis, and are transferable from one year to the next, if they go unused.  The license payment also includes a number of support hours from the vendor.

I would recommend organizations to evaluate this solution if they are looking into data extraction products. It is complex to implement and use, meaning it will require time to use effectively, but it provides good results.

We would rate this solution an eight out of ten.