We would like to implement an IDP (Intelligent Document Processing) solution within our organization. Does anybody have experience with the implementation of IDP solutions? If so, what should we be mindful of when selecting a software provider in order to be successful?
Sort by:
When choosing IDP, it is worth considering a solution with a very good graphical interface. Working with documents can be overwhelming if the interface is chaotic and difficult to use. Employees will be more effective if they have better experience using IDP.
Thank you Przemyslaw. After calls with multiple IDP providers, it is clear that user interfaces are quite different when comparing providers, so for sure this is a big differentiator.<br>One thing I'm struggling with is, is how to compare the "IDP engine" between providers ? They all claim to be the best but how to make an objective comparison of the output, knowing that each engine needs to be trained based on 100's of documents ?<br>Not sure if you have experience in this space... <br>Thank you !
Being from a background of implementor and having managed teams who implemented Capture / OCR / IDP solutions in the past, I can say with conviction that there is no "perfect" solution in the market with 100% (or nearly) accuracy.
Few observations from my experience:
1. Documents quality, including print quality, layout and structure impacts the OCR / IDP results.
2. Engage professionals to improve layouts and print quality of application forms and other standard document types, as much as possible to meet OCR standards.
3. Legacy documents should be digitized in full-text OCR pdf format, to allow full text search for users to retrieve digitized documents, since metadata capture using OCR / IDP might not be effective.
4. IDP solutions come in two forms - web-based and client-server, choose what's suitable for your organization type/structure (centralized vs distributed, operations overhead for client-server).
5. It's always a good idea to acquire solutions, which are standard-based and open in nature, e.g. extensible with SDK and provide Open APIs to integration with Content Management or ERP or CRM systems.
5. Solutions with pluggable OCR / ICR engines are always recommended, if your document types are large and complex, to address the variety of documents.
6. Many vendors might claim to offer AI / ML capabilities built-in, carefully evaluate the functioning and value of those claims.
7. Start small, with a handful of use cases, but set sight for enterprise-wise use cases and evaluate solution with wider capabilities and product roadmap, else silo IDP can be costlier than an outsourced vendor for digitization.