Using AI-powered document processing to improve the UX of digital services

Intelligent Document Processing (IDP) is a growing capability of Artificial Intelligence (AI) algorithms to convert unstructured data into structured data.

Let’s say that your company is using an expense claims app to submit, well, expense claims. You know the chore—you go through all your receipts in your wallet and enter the data into a form in the app. You add the merchant name, date of transaction, transaction amount, reasons, expense category, etc. And to top it all out, you add the picture of the receipt as proof of payment.

The receipt represents unstructured data. It has data, but you can’t process it. When you enter the claim details into a form, you are manually converting unstructured data to structured data. Well done, human worker!

If you are asking the customer to enter passport details and upload the passport as proof, aren’t you asking the customer to do the same thing twice? With IDP, the customer just needs to upload the passport.

IDP is the new digital worker in town that can help convert unstructured data into structured data. In an IDP-enabled expense claims app, you would just take a photo of the receipt. The IDP software will read the image, extract the data, fill out the form and offer it to you for review before submitting it. Awesome job, productive human worker!

IDP vs OCR

But wait a minute, isn’t the tech just plain ol’ Optical Character Recognition (OCR)? Well, OCR is just a part of the IDP play. OCR tech is good at converting data stuck in images to digital text. But that’s all OCR does. IDP tech adds by taking the digital text and applying Machine Learning algorithms to do some amazing things like grouping the data into categories, picking out names of people, places and organisations, understanding the sentiment of a statement and much more.

But wait a minute, isn’t this text wizardry called "Text Analytics" (TA) these days? You are right! Therefore, IDP = OCR + TA.

An interesting space opens up with this equation. If OCR is used to extract data from images, can’t I also use other content type extractions such as audio-to-text, video-to-text as well as unstructured digital text such as email? Yes, yes and yes. Now, you realise why IDP companies use the famous statistic: “Over 80% of all enterprise data is unstructured”. There is a lot of data in an enterprise that is unstructured and therefore un-processable in an intelligent way.

UX with IDP

What does IDP have to do with the UX of digital services?

We are seeing a growing number of digital services that ask users to upload documents. Here are five use cases we’ve come across:

1. Sign the declaration form and upload it.
2. Upload your Covid certificate.
3. Select if you want to receive the dividend in cash or shares, then sign and upload the document.
4. Enter your passport details and then upload a photocopy of your passport as evidence.
5. Take a snapshot of the medicine you are bringing, fill out the particulars and then submit the form before your flight to Singapore.

You must be wondering how IDP helps in use cases 1,2,3. Yes, IDP has no impact on the customer experience here (we might argue and say that for use case 1, we might use the likes of DocuSign), but IDP helps greatly in backend operations.

The questions for backend operations are:

• Is the document signed? When was it signed?
• Is the Covid certificate legit? When were the shots administered? Where?
• Which option is selected and is the document signed?

As you can see, backend operations have many questions that today is answered by human workers. With IDP, these questions can easily be answered by digital workers (a bot, for example).

You can now see how the UX of the admin flows will change when using IDP. Instead of designing to show data, you now have to design to include outliers, fakes, and suspicious activity actions—wonderful benefits of working with structured data. With IDP you are designing for different outcomes.

In use cases 4 and 5, IDP helps improve the customer experience as well as the backend operations.

If you are asking the customer to enter passport details and upload the passport as proof, aren’t you asking the customer to do the same thing twice? With IDP, the customer just needs to upload the passport. The software extracts the relevant details and offers a preview which the customer can use to review the details and make changes where relevant.

The backend operations also benefit from such a process. Instead of matching all passports to details submitted, backend operations need to only focus on where the customer changed the details and where the IDP software signalled low confidence understanding the data (yes, they can do that as well).

The above use cases also show how the UX of the digital service (both frontend and backend) will change. With IDP, the customer does not have to upload a document or enter a lengthy form. They can just use their camera to take a photo of the document and review the details.

Summary

Digital services are only going to grow as more and more of our activities and tasks shift online. The overall UX is going to be the big differentiator. IDP offers organisations to way to leapfrog the competition with faster, frictionless digital services.

Contact us to see a demo of how IDP software works.