What is our primary use case?
We have been using UiPath IXP previously after utilizing UiPath's Document Understanding for one of our clients. We found that significant improvements have occurred with UiPath IXP, which is an intelligence extraction process that uses LLMs. The use case is for a manufacturing company, and we have used this for the extraction of POs and non-POs. It is a large workflow that we have automated. Incoming every month are around 1000 plus invoices. We extract the information and send it to SAP, and from SAP, other processes run. We have utilized UiPath IXP for this extraction.
There are different kinds of invoices we use for a supply chain process with ERP and vendors in the manufacturing company. Whenever they purchase something from the vendors for the manufacturing unit, diverse invoices will come. We have used those invoices with UiPath IXP, where we have mapped those fields. The model has categorized and classified all the invoice types and facilitated the mapping of the fields, taxonomy, and the mapping of the result confidence. After this extraction is completed, we take all those fields and inject them into SAP so that the next process, which is a large workflow, occurs smoothly. UiPath IXP has to identify a specific kind of invoice and needs to extract those fields with a sufficient level of confidence, which it successfully does before injecting to SAP.
What is most valuable?
The best features UiPath IXP offers include its model training capability, which we have been using previously in Document Understanding. We had invoice models available and would manually map the fields. We needed to run the model, run the pipeline, and then use that skill. With that skill, the classification and extraction occurred, requiring us to upload data sets, invoices, and more. When it comes to UiPath IXP, we just need to inject for the classification and extraction of the same set of invoices, and afterward, the mapping happens automatically. Due to the training imposed on the UiPath IXP model, once that mapping has happened, there is an opportunity for a human to check the fields, although usually, most of the fields are mapped correctly. User intervention is limited. Once the mapping is done, the model will be trained upon those things, taking significantly less time than in UiPath Document Understanding. All these things happen within a matter of time, which I would say is the best feature.
Both aspects are significant because regarding user intervention, if you want to map different sets of invoices and if you are mapping it every time, it takes too much effort that is unnecessary, which UiPath IXP is reducing. Additionally, whenever you map some fields and upload a data set to train or evaluate a pipeline, these manual steps become unnecessary for the developer. You just need to understand and map it. The training and everything happen with UiPath IXP in a notably shorter time because it is a pre-trained model that already knows everything. Both aspects hold equal importance.
UiPath IXP has positively impacted our organization by reducing the processing time while enhancing the model capabilities and the diversified nature of the model. We can inject anything without needing to find the kind of input we are feeding it, as it automatically identifies and classifies the type of invoice. Additionally, the extraction capability, speed of extraction, and ease of use have improved. When discussing models and pipelines, it involves less effort than before, making it easier to use.
We have seen a measurable difference in processing time with UiPath IXP. Previously, the processing of 1000 plus invoices would require six to seven FTEs working for 10 to 15 days. Now, we accomplish this in a couple of days with UiPath IXP due to the faster processing time. We utilized different machines, about three to four, giving queues for availability. Once the machine is available, the processing of the invoices with UiPath IXP occurs much better and faster.
What needs improvement?
When it comes to improvements in UiPath IXP, there is a challenge with different languages, particularly Arabic invoices or handwritten ones, where there is some mismatch or difference in extraction. This could be improved. For instance, if there is a "Z" available on the screen, it may be identified as "2." Improvements in these two areas from the extraction point of view are necessary, along with considerations regarding pricing levels.
In terms of support, UiPath is performing very well. Whenever we require assistance from UiPath, it has been positive. However, I would suggest that there is room for improvement in identifying complex different layout invoices, particularly for handwritten text, since there is potential for improvement in that area.
For how long have I used the solution?
We have been using UiPath IXP for a significant amount of time.
What do I think about the stability of the solution?
What do I think about the scalability of the solution?
UiPath IXP is pretty scalable and works effectively.
How are customer service and support?
UiPath is performing very well in terms of support. Whenever we require assistance from UiPath, it has been positive. I would rate the customer support at 10 out of 10 because UiPath has maintained excellent customer support and community engagement for many years.
Which solution did I use previously and why did I switch?
We previously used UiPath's Document Understanding for a different client. At that time, there was no model training, which has since scaled up and improved through UiPath. We moved from one stage to another but did not switch from a different provider.
What was our ROI?
Observing a return on investment is clear. UiPath IXP performs better for organizations with a high inflow of invoices, and when you couple UiPath IXP with automation, it significantly increases ROI. The UiPath IXP model charges a little higher, but when bundled with automation, it generates a good ROI. It is best suited for large-scale companies, handling around 1000 plus invoices in a month or even up to one lakh invoices.
What's my experience with pricing, setup cost, and licensing?
My experience with pricing, setup cost, and licensing has been somewhat complex. I feel that the pricing structure is a bit high.
Which other solutions did I evaluate?
We have evaluated other options such as different OCRs, including Azure OCR.
What other advice do I have?
My advice to others considering using UiPath IXP is that if you have a large set of invoices or any kind of documents needing extraction and if you are already using UiPath, it will be a better solution for OCR extraction, bringing significant ROI if you have a substantial number of documents incoming. If you couple this with some kind of process for automation, it will yield considerable ROI and ease in working. UiPath IXP is a pretty solid model that delivers good ROI to clients and reduces developer effort. Compared with Document Understanding, it is a better model. I rate this review a 9 out of 10.
Which deployment model are you using for this solution?
Public Cloud
If public cloud, private cloud, or hybrid cloud, which cloud provider do you use?
Other