Skip to main content

I have been thinking about this for a while. How is the Document AI problem that we are solving is quite similar to pure system integration problems? Let us say you want your Procore to synch with your Foundation. In that case you will have an integrator connect the two sql databases and ensure that the project management system like Procore talks to an accounting system like Foundation. In this scenario of system integration the same company owns the two databases (Procore and Foundation) so a system integrator can build the integration without having to deal with data access issues.

Lets say you as a construction company just won a bid and now its time to execute. You get thousands of quotes from various vendors. How do you get these quotes and create POs from these quotes efficiently. The quotes information is going to be residing with those thousands of vendors – some have ERPs and may be even some are generating this quote information using Excel. The construction company can not go and ask all these thousands of vendors that are sending these quotes for their database access so the information from quote in vendor’s database can flow to construction company’s database. To solve this problem using pure sql db integration is infeasible because of the combinatorial explosion in number of integrations you need to build.

This is the type of problem where Document AI shines. The vendor quote information from vendor’s database gets put into a quote document (usually pdf) that the Document AI parses and converts it to a structured form. Now this structured information can get passed to the construction company’s database. Essentially instead of requiring thousands of system integrations to be built for this construction company the Document AI cuts the number of integrations to just 2. There is a huge amount of time waste in document processing and I hope one day the industry will realize amount of time that can be saved in document processing with Document AI.

Leave a Reply