UiPath Document Understanding Reviews

We are a service-based company, having a team size of 120 RPA developers.

The core technology of UiPath Document Understanding is OCR, which can be either ABBYY or Google Vision. The drag-and-drop facility, decision-making accuracy, and user-friendly actions make UiPath Document Understanding helpful for users.

UiPath Document Understanding offers a feature that allows users to process specific documents, such as purchase orders or invoices. The tool provides the ability to define and segregate actions and highlight fields using something similar to rubber band techniques. This means, in an invoice, users can easily highlight specific fields within the document and train the model accordingly. The process is user-friendly and can be easily executed even by those with basic knowledge of flow.

One area of improvement for UiPath Document Understanding is the accuracy of handwritten documents. While the solution supports handwritten documents, the accuracy percentage is not as high as desired. Therefore, it would be preferable if UiPath could focus on improving the accuracy of handwritten documents in their next release. Although they currently support handwritten documents, the accuracy is still low.

Also, the documents for the solution are lengthy, and because of this, I say that the solution should incorporate some video sessions.

I have experience with UiPath Document Understanding, in which the users are provided with multiple types of Document Understanding and a process-wise range of available documents. Overall, I have been using the solution for three to five years. Also, I am using the latest version.

The solution is quite stable, and I am happy with its performance in my programs and processes. I would rate the solution's stability an eight out of ten

The solution's scalability with the cloud version is really good. I rate the solution's scalability an eight out of ten.

Around 30-40 people work on UiPath Document Understanding in our company. The extent to which we use the solution, particularly in terms of utilization of bots or document processing, is dependent upon the volume of usage. In other words, if there is an increase in the number of customers, we will likely increase the usage of the solution.

UiPath's technical support is good. Even in emails, they are responsive and helpful. They also have a very good library of troubleshooting documents and other resources. Overall, I am happy with the solution's technical support.

I have experience with Automation Anywhere. Also, UiPath Process Mining is a pretty good tool since they have documents which provide its users with a clear understanding. On top of that, they also provide videos. So it's quite easy for an end-user to understand.

Although the solution's initial setup is not overly complex, UiPath could work to make it more understandable for a layperson. They could improve the setup process by creating a more structured flow.

The time needed for deployment varies, depending on factors like the use of cloud services. With a Plug and Play model, UiPath provides cloud services and a link, after which we can begin using the solution. However, deploying on a private cloud may take longer, up to three to five days or even two weeks, depending on the hardware readiness.

In terms of the maintenance of the solution, let's consider a scenario where we need to initiate a programming project to develop a bot that can help in invoice transactions. Typically, this process can take anywhere from three to five weeks and requires at least one project manager who oversees a team member who understands UiPath and business processes. To successfully develop the bot, we need to first analyze and understand the relevant business processes. Once we have derived the business processes, we need to document the solution, including creating a PDD. This document outlines the process and how it will appear once automated. Apart from that, you should have a good set of developers who understand UiPath.

In our organization, we implemented the solution with the help of integrators.

The ROI from using UiPath Document Understanding is really good, and we currently see a return on investment of around 40% to 50%.

For the licensing part, in our company, we initially started off with a monthly plan, but yearly plans are also there. Cost-wise, UiPath  Document Understanding is good since it provides flexibility in its pricing. So, smaller firms, especially MSMEs, can opt for monthly or yearly plans based on their budget. As an end user, I wish they reduce the price of their pro version to 300 dollars since UiPath Pro starts somewhere around 400 dollars per month. UiPath's competitor, Automation Anywhere, offers a plan for 99 dollars per month.

I can definitely recommend the solution to other users. Overall, I rate this product somewhere around eight out of ten.

UiPath Document Understanding allows us to extract information from documents, classify the document type, and optionally validate the data with a human. We also train the bot to learn from the mistakes it makes.

We extract data from boxes and tables, as well as handwritten words.

UiPath Document Understanding helps us extract from certificates, the names, addresses, states, dates, taxpayer identification numbers, and signatures. We also extract information from health insurance documents such as the name, patient's address, code, and contact information. Additionally, we have invoices that we have to extract from a table with rows.

UiPath Document Understandings' ability to handle different document formats depends on which extraction tool we use and the clarity of the documentation.

I am very satisfied with UiPath Document Understanding AI and machine learning.

UiPath Document Understanding integrates well with other tools. Some of my colleagues have integrated the solution with ABBYY.

It has helped reduce human error.

The most valuable feature is the ability for humans to train the bot. It can learn from human experience, so if we include steps in our framework for humans to validate extraction or classification, the bot can learn from them to improve document understanding and precision in future document processing.

The extraction and classification process should be simplified.

I have been using UiPath Document Understanding for six months.

UiPath Document Understanding is stable.

UiPath Document Understanding is scalable.

The technical support is good.

Positive

The initial deployment was straightforward. The deployment took three hours to complete.

I implemented the UiPath Document Understanding myself.

I would rate UiPath Document Understanding eight out of ten.

Our organization also uses ABBYY.

We saw the time to value within the first three days of using UiPath Document Understanding.

We use UiPath to process HR documents. It helps us with the last mile of the paperwork. We're dealing with a lot of offline legal documents that need to be converted to Word or PDF. We use UiPath to read the documents and process them into a digital form the company can use.

We're consultants, so we have to provide invoices for our services. We have a specific layout that's internally defined, and we store the information in SAP. It's processing thousands of documents. Most of them are PDF forms. Our customers and partners usually save these documents and send them to us by mail or email. Almost all of our documents are processed through UiPath.  

UiPath has saved us a lot of time because 90 percent of our document processing is now automated. I can't give an estimate of how much we have reduced human error or the number of resources needed, but I can say that we are using time more efficiently. The solution makes work easier for our staff and frees up time. When we automate most of the documents, staff has more time to focus on issues in the remaining 10 percent. 

I like UiPath's machine-learning capabilities. My first impressions were good, but we only use it in a few processes. We intend to expand our use of these features in other situations, but we're still learning how to use the AI functions. We use the OCR much more. 

You can use machine learning with OCR to read and interpret documents. You don't need to map out specific fields because machine learning predicts where in the form the data is expected. Machine learning can identify various types of data in different sectors. UiPath is easy to integrate. We have only integrated the solution with two systems so far, but it's working well. 

Sometimes we have challenges when we need to read something like a barcode, so we must use the Cisco algorithm to solve the issue, or I have to ask developers for help specifically to capture this kind of information. There are other processes, such as refunds, that we still must do manually because there is no way to use automation to solve this issue. 

We have used UiPath Document Understanding for nearly two years.

I rate UiPath's support an eight out of ten. When we call their support team, they always solve our problems. That's what matters. 

Positive

I rate UiPath Document Understanding an eight out of ten. It's an excellent product that's stable and feature-rich. We also have the option to augment it with some external features. 

My company has a confidential website where timesheets, salary statements, and other information regarding employee working hours are provided in PDF format. I plan to use UiPath's Doughnut Scatter and Keyword Classifier to extract specific data from these PDF documents.

The best feature, in my opinion, is, we can train the document using a machine-learning classifier. For example, if the PDF documents are of different sizes and formats, we can train all six templates with the Keyword Classifier and the Template. 

Currently, we have to train multiple templates because the column size and row size change in each PDF. My solution is to improve the process by training only one particular template so that the information can be extracted regardless of the row size and column size variations in the PDF. This way, we won't need to train each and every template.

I have been using the solution for two months.

The solution is stable.

The solution is scalable and this is the main reason we are using UiPath in our company.

The initial setup is straightforward and easy to understand.

I give the solution a nine out of ten.

We have a team of six developers that use the solution.

I recommend the solution. UiPath has recently added OCR capabilities to its Document Understanding feature, previously unavailable. This new feature is called UiPath OCR.

Our use cases are in the banking sector. We have certain PDFs that contain customer data, and we have to extract data from them and compare it with the data provided earlier. The PDFs can be scanned, handwritten, or native PDFs.

It gives us a complete, unattended process for extracting details from PDFs. Otherwise, that would been redundant work done by specialists. It is saving us around 800 workers.

For one record, we used to take about half an hour, but Document Understanding completes things in seconds. It's very quick. It has also reduced human error a lot.

In UiPath Document Understanding, I use two extractors. One is the Intelligent Form Extractor and the other is the Machine Learning Extractor. The first one is very useful and user-friendly. When we have a single template or two to three templates, Intelligent Form Extractor is much easier to use. But if we have multiple templates, it's better to go with ML Extractor as you can train the model with different templates.

The artificial intelligence features are very good. The more you train it, the more you will benefit from it.

Also, integrating Document Understanding with other systems or applications is not complex. It's quite easy. Even for attended processes, if you combine UiPath Document Understanding with Action Center, it's almost like unattended processing.

Document Understanding is a combination of the extractor and OCR. Depending on the taxonomy that you give and the classification that you define, it is accurate. But there are multiple OCRs and I find UiPath Document Understanding to be as accurate as Microsoft or Textract OCR. If the accuracy was improved, it would be much better.

There are also problems with handwritten documents. The results from those are not 90 percent accurate. With scanned and native PDFs, it is 90 to 99 percent accurate, but the accuracy of handwritten documents is somewhat less. If they could improve on that, it would also be helpful.

Also, in the banking sector, we can't use the cloud-based Orchestrator or cloud endpoints for OCR. Getting approval for ML Extractor takes a lot of time for us. Instead, if we could go with Intelligent Form Extractor, it would save a lot of approval time. If they could improve and add features to Intelligent Form Extractor, it would make our lives easier.

I have been working with UiPath for five and a half years and with Document Understanding for one and a half years.

I have not experienced it crashing or lagging.

The scalability is good. I haven't seen any waste of money.

We have not contacted their technical support very much.

We do not use UiPath in the cloud; we have our own internal version, so the initial setup took time for us. But that had nothing to do with UiPath. It was because of our company's policies.

It took time for people to understand it, but once they got it, it went well.

It took some time because Document Understanding is pricey compared to other tools in UiPath. After 10 to 15 days, when there were fewer errors, that's when we started to see profits.

I would suggest combining other technologies with UiPath. That's when you see more benefit from Document Understanding.

The product is very, very nice. The tool relieves the data entry team from manipulating data. It shortens the time of extracting data by 70 to 80%.

UiPath should localize the built-in features for supporting Arabic scripts so we do not rely on third-party tools. It will take some investment. Sometimes, when we use third-party tools, we need to collect multiple samples of the same document in Arabic. It creates some errors in understanding. 

Sometimes, we need to do multiple formats for the same document. It happens due to the nature of the Arabic language. For the same document in English, we only supply one template. The output of handwritten documents in Arabic is very poor, regardless of the solution. For computer-printed documents, we need to tune the system.

I am using the solution currently.

The product is stable.

The tool is scalable.

We were using our own intelligent business processing solution. It is more convenient to use because it's our own product, and we rely on Amazon Web Services. It provides big advantages for Arabic scripts. It is more understandable.

The product is easy to implement to an extent. If we are familiar with the input, processing, output, and how to declare variables, it's easy. If we are new to UiPath, it would be somewhat difficult to understand. UiPath Academy and certifications are very important.

The difficulties in implementation depend on the environments, document sources, document types, and customer understanding. Sometimes, customers think the product can pass anything without configuration, which is not true in some cases. We have to be very clear with the customer about the type of documents and layout they can expect. Two working days are sufficient for configuring, extracting, and linking the process. One person is enough to deploy the solution.

The pricing depends on the context of the project. UiPath is a pricey tool for small customers. If the customer agrees on the cloud product, we discuss the cost accordingly. In some cases, even though the customer sees value in the product, return on investment, productivity, and enhancements of their processes, they decide not to choose UiPath due to budget constraints. They choose a different vendor.

Sometimes, the customer is small, but we could see the potential for using the tool because they might have multiple processes. The price we offer is based on the context and the size of the opportunity we get for references. Document Understanding has helped us automate processes like contract management, reconciliation of invoices against purchase orders, claims management, and HR processes.

We process 100,000 to 500,000 documents in a year. We also had a project for Dubai Customs to reconcile the customs clearance, which involved 500,000 to 1,000,000 documents. These documents contain tables, bar codes, dates, and checkboxes. Tables might span over multiple pages, and we must capture it completely. It becomes challenging sometimes. If there is an invoice of three pages, everything must be captured, but sometimes the values are inaccurate.

Around 70% of our organization’s documents are completely processed automatically. We haven't used the product for signatures. Customers often see from a productivity point of view. If their employees process ten documents an hour, and the tool processes 50 documents an hour, using the tool is an advantage for them.

The tool does not work for integrations. It extracts data. When we extract data successfully, we rely on other business tools for integration. Document Understanding serves as a first milestone in the process. The tool at the second milestone would pick up the extracted data and post it according to the processes or applications used.

We need human validation of Document Understanding outputs three out of ten times. It does not take more than 20 to 30 seconds per document. An employee would take ten documents per minute. RPA could handle 30 to 40 documents per minute. Document Understanding has helped us reduce human errors by 70% to 80%.

Document Understanding has helped free up staff’s time for other projects. From a development point of view, it has freed up 70% of staff's time. From an employee's point of view, it has freed up 80 to 90% of their time. We saw the value of the product after a couple of projects, especially when we implemented it and saw the value for ourselves. As a customer, vendor, or implementer, when we see the execution, we see the value.

If someone wants to take advantage of this technology, they must think multiple times about different scenarios. They must centralize the capture or the source of data. They can use any product that is easy to use and configure and link it to the processes. They must sync the templates and the configuration multiple times and link it to the automation strategy.

Overall, I rate the solution a five out of ten.

We use UiPath Document Understanding to extract data from invoices so that the data can be entered into SAP applications. We often receive multiple invoices in different formats. To use UiPath Document Understanding, we need to train the document. This means that we need to provide a sample template that the robot can use to learn how to read the invoices and extract the desired data. The labels in the fields are important, as they tell the robot what information to extract. 

For example, the invoice number, seller's address, quantity, and any other required information. The table of line items also needs to be extracted. We define the fields from which we need to extract data. We use the form-based extractor or machine learning extractor to extract the data. We export the extracted data to an Excel file. This allows us to collect all of the data from multiple invoices in a single location. We can also implement a validation step to ensure that the robot has extracted the data correctly. This can be done easily by using the human in the loop to manually review the extracted data before it is entered into SAP applications.

Our organization has many invoices of varying formats. To simplify the process, we implemented UiPath Document Understanding to extract the required data from all the unstructured invoices and import it into a single, structured Excel sheet.

Using the different endpoints and machine learning, we can extract data in different formats including handwriting and signatures.

UiPath Document Understanding's machine-learning capabilities are good. We use it to extract the information we need from documents of any format, and then align it with the corresponding table.

We easily integrated UiPath Document Understanding with UiPath Studio by connecting the data flow from the different applications to extract the necessary records.

We have so many invoices that UiPath Document Understanding helps us classify them. The document is first classified into a type of document, such as an invoice or a resume. This segregation of data allows us to send the invoice templates to our folder and extract the data and all the details without having to read them one by one.

With UiPath Document Understanding, less and less human validation is required over time. Each time we run the process, we can select from an option for attended or unattended automation. The other option is to use the confidence score to determine whether human validation is required.

The handling time before automation for one PDF document was five minutes and now with automation, it is one to two minutes.

UiPath Document Understanding helps to reduce human error by removing the fatigue factor and by continuing to run after hours without breaks.

There are many features that can be added to the action center to keep humans in the loop to evaluate the accuracy of the extracted data. We can also set a confidence score threshold, such that if the confidence score is greater than 80 percent, we can take a certain action. Otherwise, we need to manually correct the problem in the action center.

UiPath Document Understanding could be more user-friendly. There are so many endpoints, and entering the API is also a manual task. Currently, it is a long and complicated process with many steps. If we remove all of those steps and use the ChatGPT or OpenAI library, we can start using the solution with fewer steps.

I have been using UiPath Document Understanding for six months.

UiPath Document Understanding is stable.

UiPath Document Understanding can be scaled for multiple invoices.

The deployment was straightforward and required a few days.

I would rate UiPath Document Understanding eight out of ten.

We are processing around 12 documents using UiPath Document Understanding. The documents are in PDF format.

We have four members using the solution.

My use case for UiPath Document Understanding is invoice processing, as most vendors share invoices that need to be posted in the ERP (Enterprise Resource Planning) software. The vendor sends invoices via email whenever my organization purchases goods or products, so the ERP is updated whenever a product is received. Still, before making a payment, there needs to be some validation. For example, you have to check that the invoice sent matches the product received, so this is where I use UiPath Document Understanding.

I use the solution to extract the data from the invoice. Then I validate it with the ERP software in terms of which product has been received, including the unit price, shipping charge if applicable, the tax, which depends on the state, and product quantity. My organization needs to validate all data before paying the vendor, which is where UiPath Document Understanding becomes useful.

The most common type of invoice processed within the organization is the purchase order, which is more structured, so extracting the data is easier. It's unlike a utility bill, such as an Internet or electricity bill, with no purchase order and a different invoice structure. It makes it a bit more challenging to train the machine learning model in UiPath Document Understanding when processing utility bills versus purchase orders, though not impossible.

Currently, my organization focuses on purchase orders in terms of UiPath Document Understanding usage. Still, once that's successful, there is a possibility of using the solution on other invoice types.

About sixty-five to seventy percent of the invoices are automatically processed with the help of UiPath Document Understanding, though it could vary from business to business. Failures rarely happen and are usually the result of the business rules or conditions. About twenty percent of the processes require human intervention.

UiPath Document Understanding helped improve my organization by reducing human errors and intervention. Humans handled the entire task before the organization used UiPath Document Understanding, and it was a huge task that could only be completed within the year. In contrast, UiPath Document Understanding makes it possible to complete the task within a month, specifically completing all invoices that flow monthly within the same month. In the past, you had to delegate more work to humans, but after integrating with UiPath Document Understanding, the task required less human work, and the solution used up fewer resources compared to the amount of resources used up by humans doing the job. The solution enabled humans to work on other tasks, helping free up staff time by sixty to seventy percent.

I found UiPath Document Understanding helpful in automating different types of processes. One process can be split into multiple processes or sub-processes.

For example, when the vendor sends emails, I can download all emails through automation, mainly via UiPath Document Understanding. Downloading the emails can be a continuous process, scheduled at different intervals in a day. The next sub-process made possible by the solution is extracting the data from each invoice.

You can also use UiPath Document Understanding for ID management or extracting data from IDs, such as passports or government-issued IDs, which I like. 

Signature matching is another feature of the solution that you can use, which I found good, though I only tested it. That feature wasn't used in production because, in my organization, UiPath Document Understanding is primarily used for data extraction from invoices.

I also found the parking feature in the solution interesting. For example, when you train the model but have data that's not extracted completely, or you are unsure, you can park that invoice and then have a human interfere to validate it.

UiPath Document Understanding can handle varying document formats, such as handwriting and signatures. Still, accuracy in data extraction from handwriting is fifty-fifty and would depend on the quality of handwriting. For example, the digit seven could be mistaken as one. For signatures, about sixty percent of the data matches or is correctly extracted by the solution.

The AI or machine learning capability of UiPath Document Understanding works and is successful seventy to eighty percent of the time. My organization trains the model with a set of invoices from each vendor to pick specific data, such as invoice number and date. Data extraction success varies with the number of invoices trained and if there's a new template, but you can expect the success rate of the machine learning module in UiPath Document Understanding to be between seventy to eighty percent.

UiPath Document Understanding has two parts: the AI, similar to integrating ChatGPT to your API, and the forms, where a human can validate the form as the solution acts as a parking station where the data extracted by the bot, if you're unsure of the data or the data wasn't extracted properly, may be parked there so a human can intervene and validate it.

In terms of how often human validation is required for UiPath Document Understanding outputs, you'd put specific logical conditions after the extraction. For example, the quantity, unit, and total amount would be integrated. Suppose the quantity and unit price are extracted correctly, but the total amount was incorrectly extracted. In that case, that won't always need to go through human validation because UiPath Document Understanding can automatically calculate. For one hundred invoices, at most, five to six invoices go through human validation, while the rest don't have to.

UiPath Document Understanding can be integrated with other systems and applications, and my organization has tried it by integrating the solution with Python. My organization built a machine learning model with another system where UiPath Document Understanding was integrated with UiPath data services, and there was even a dashboard that gave insights on how many invoices were processed, which invoices fell under which state, and so on.

An area for improvement in UiPath Document Understanding is data extraction from scanned documents. It could be better because I've observed a ten percent failure or flaw whenever the vendor scans and converts documents into PDFs.

I've been using UiPath Document Understanding for two and a half years.

Sometimes, the bot or system resources cause UiPath Document Understanding to crash, but not frequently, and it's covered by the one-hour maintenance a month. I find the solution stable.

Scalability in UiPath Document Understanding is very good because my organization can scale the same logic or script it developed with UiPath Document Understanding into any of its systems. If you deploy the solution in one ERP system, for example, you can install it in another system at any point in time, and configuration would take about one to two days, plus testing. UiPath Document Understanding is scalable.

I've contacted the technical support for UiPath Document Understanding. At the initial deployment stage, a specific resource was allocated to my organization, mainly because the solution was new. It was in the beta stage so that I could contact the resource for any difficulties or doubts, and I've contacted the resource multiple times.

I'd rate the UiPath Document Understanding support team a nine out of ten because sometimes there was a delay in communication as the previous resource was from a different time zone, so another resource was allocated to my organization, which belonged to the same time zone as mine.

Positive

My organization used to do the tasks manually and tried Power Automate for automation, but found that solution a bit difficult compared to UiPath Document Understanding. UiPath introduced UiPath Document Understanding to my organization, and the organization found that it was a proper solution.

I was involved in the initial deployment of UiPath Document Understanding, and I found the process straightforward. You just had to activate the licenses one time. After that initial license activation, you can quote the number of licenses you need to UiPath, and it will be automatically updated in the Orchestrator. The setup would take around one week to fifteen days in the initial stage, and the maintenance comes after, which takes just one to one and a half hours.

I deployed UiPath Document Understanding with about three to four team members. The solution was deployed in-house.

I've seen ROI from UiPath Document Understanding for the past two and a half years because the solution helped reduce about seventy to eighty percent of human tasks, which meant that my organization could utilize the staff in other processes that could not be automated or tasks that had non-standardized logic or conditions. Though the ROI could also depend on how much you invested and your actual staff, I've observed ROI from UiPath Document Understanding.

My organization saw the potential in the solution after some demos but could not see the value immediately. Initially, the results weren't as good, with some difficulties. Still, with the help of the UiPath support team, improvised data extraction, and two to three months after moving UiPath Document Understanding to a production environment, the organization saw value from it.

UiPath Document Understanding offers fair pricing, primarily if you consider the number of resources and human work the solution reduces, plus the ROI you can get from it.

My organization had a native PDF document extraction tool that could have worked if the vendors had similar invoices or templates, but that wasn't the case. My organization also tried a machine learning tool with some outsourced Python, but it wasn't very standardized or robust, so it wasn't good to use versus UiPath Document Understanding.

UiPath Document Understanding has a public cloud deployment within my organization, as there was a constraint with deploying it on-premises, requiring a VPN within the same network. Hence, my organization chose to install it on the cloud via a cloud orchestrator that even allowed access from anywhere in the world.

The initial deployment of UiPath Document Understanding required two people, but as the number of businesses increased, my organization had to add five to six people to the deployment team to handle the installation, script maintenance, and connecting to the client center. The installation requires three to four people and two more for support. Six people to deploy UiPath Document Understanding for thirty to forty businesses would be sufficient.

UiPath Document Understanding doesn't require much maintenance because UiPath usually does automatic updates or feature updates to the cloud orchestrator and sometimes to the scripts my organization built. Maintenance is minimal and would only take about one hour per month.

My rating for UiPath Document Understanding is eight out of ten because it still needs improvement in the quality of scanned documents and data extraction.

My advice if you're looking into UiPath Document Understanding is to check whether it's a solution you really need. Think about the tasks requiring UiPath Document Understanding and the features you're looking for. Consider the costs and time associated with the solution, as the clock would start ticking after you buy the license. You need to analyze precisely why you need UiPath Document Understanding and what features you need from it. The team responsible for it should also be well aware of the solution's features to maximize its use.

My organization is a UiPath partner which buys the license and provides it to businesses.