What is our primary use case?
The major use case I built using
UiPath AI Center was for automated invoice extraction, where I created an ML model that extracts all the required fields from invoices. This helped me automate the process of invoice processing in SAP, and I used the default ML models provided by UiPath to train and fine-tune them according to my requirements. I trained approximately 1,000 documents in
UiPath AI Center for this automation.
Training and fine-tuning those ML models for invoice extraction was straightforward because UiPath AI Center provides a very good interface, allowing me to use it without needing deep technical knowledge. I simply upload some invoices and manually mark them using the provided markers; the interface is very friendly and easy to use, enabling any new person to quickly catch up with it. There was no significant learning curve involved.
Regarding my main use case and my experience with document intelligence and invoice extraction automation, UiPath AI Center gives me everything I need. For instance, it also provides Action Center, from which I can train invoices with lower confidence. I set up a pipeline that retrains these low-confidence invoices automatically in UiPath AI Center, which has helped establish an automated pipeline process for training ML models without my intervention.
What is most valuable?
UiPath AI Center offers several great features, including the ability to create datasets manually and automate the pipeline for retraining ML models. I can easily set a default timing or create a recurring pipeline to train the documents from Action Center, which provides me with a variety of features, with the easy user interface being one of the best aspects.
The recurring pipeline scheduling has saved me a lot of time because I usually have to manually train and mark documents, but when users utilize my bot, they contribute to the data, marking approximately 50% of invoice extraction fields from Action Center. This allows the existing marked data to be automatically retrained, significantly saving me time as a developer and improving the training efficiency since multiple users are validating the invoices, generating a lot of data for training.
UiPath AI Center has positively impacted my organization, particularly in automating the invoice extraction process, leading to significant cost savings and time reduction. Manually reading and posting invoices in SAP involves considerable time and effort, but by automating invoice extraction and posting, we now only require users to flag low-confidence fields, allowing for major cycle time reduction and cost savings in invoice processing.
I estimate a 50 to 70% reduction in cycle time and roughly a 60% reduction in the required manual effort. Some humans still need to validate the extracted data, so it may be around 60 to 75%.
What needs improvement?
One improvement I would like to see in UiPath AI Center is the addition of more ML models. Although it currently offers a variety, integrating with
Hugging Face to access and deploy their models in our UiPath AI Center ecosystem would enhance its capabilities significantly.
For how long have I used the solution?
I have been using UiPath AI Center for two years, and it was the first product I worked on, where I built many automations and bots using UiPath platform. Specifically, I have used UiPath AI Center to build document intelligence.
What do I think about the stability of the solution?
There have been no issues with scalability in UiPath AI Center, and the UiPath support team is exceptional in helping us manage any requirements that arise.
How are customer service and support?
My experiences with customer support have been memorable; there have been significant issues since we are using it on-premises, and when I raised customer support requests with UiPath, they resolved them quickly—often within one to two days.
Which solution did I use previously and why did I switch?
I have not exactly used a different solution for invoice extraction or automation, but I track other document intelligence platforms including
Azure and
Power Automate. However, we primarily use UiPath AI Center as it works effectively for us.
Before selecting UiPath AI Center, we evaluated Azure Document Intelligence, but this proved to be a better option in terms of cost and the fine-tuning capabilities it provides.
What other advice do I have?
The variety of ML models provided is impressive, along with customizations for different types of invoices, such as those specific to India or generic invoices, as well as receipts. This is definitely a plus point.
My advice for others considering UiPath AI Center is that it has been a great option over the past few years, but in the current landscape of agentic AI, there are many other choices available. Cost-wise, several open-source options exist, although UiPath AI Center remains a convenient solution for now. I would rate this product a 9.
Which deployment model are you using for this solution?
On-premises
