Here we're going to look at an income validation workflow that could be part of a KYC process. In this instance, possibly mortgage package submission. And what we're going to be doing is the usual Hyperscience automation around classification and identification for semi-structured and transcription, all at our high level of accuracy or accuracy target with straight-through processing On top of that, what we're going to be doing is looking to validate those documents.

So essentially we'll be taking the documents in as part of the whole submission validating the end of year payment summaries along with the income that's been stated on the application form.

We're also going to be validating the bank statements to make sure that within the date range specified, as well as validating the payslips that they're sequential and within the last couple of months as well.

So when we have a look at the documents that we're throwing through, we can see the application form that we have here, noting that we know which field were, validating the income that's been specified.

We can see on the second page we've got one of the pages is put in upside down and we can see how that auto rotates that and then so on and so forth on that document.

What we'll be comparing it to is the end of financial year, end of statement, pay as you go summary here where the income has been stated for 2019 and also slated for 2020. On top of that we mentioned we're looking at the individual payslips are they within the period that has been specified for one, two payslips and also the bank statement over here and the bank statement over here.

So when we look at the results what we find when stepping through the actual submission itself we now have within the submission looking at the form the complete extraction of the document.

We can see that the numbers have been extracted out in the names as required and all within hyper sciences human intent based recognition where it knows what to ignore using drop out and other factors.

We can see we've extracted eight 33 there from the currency amount and in fact you can see the second page is also rotated quite neatly there when we have a look at the forming task which allows us to see the validation points.

So now we're looking at the validation points from the documents that have been confirmed we're now able to see these specific validation points here. So when we're looking at the reference, the years of validation between 19 and 20, we're looking at the, the amount, the dollar amount that's been stated.

Here what we're looking for is from the income that has been produced on the end of year forms, we can see that what we're looking for is a maximum allowance of income variance of 5%.

So we can make sure that what's been stated is within a 5% variance of the actual payslips. We can say that we've found the bank savings and start dates are such and as well as the payslips within range.

So this allows us to have easy decisions based on the machine giving us the recommendation on what's been found and where the user can now successfully collate and review and confirm that we have all the dates that are valid.

Stated income is invalid, the month's invalid, and that all the payslips are present. And this is just a simple matter of how Hyperscience can work with validation amongst it's very accurate, or hitting the accuracy target on transcription.