Project ACE - FAQ's

#

FAQ

Answer

1

What is the early adopter program?

It’s a chance to get hands on with Rosslyn’s automated classification engine, powered by AI.

Users can select line items and have AI auto classify them against your taxonomy, provide feedback on the categorisation and test the functionality against different scenarios.

2

What are the goals of the program?

1. To give users the ability to automatically classify data using some of the latest and greatest technologies (Generative AI)
2. To obtain valuable feedback from users on the current feature set and future enhancements
3. To gain valuable feedback on the accuracy of classification, across a wide range of taxonomy structures, industries and data structures

3

How can I sign up?

Speak with your CSM, who will provide you with visibility of the commercials and enable the feature for you.

4

What will the functionality be?

You’ll be able to send off unclassified lines to be classified against your taxonomy structure. The system will use descriptive content found in the invoice, PO, chart of accounts, material groups etc to help the system understand what the invoice line is for.

The system will auto classify the data, and provide visibility of the chosen category, an explanation behind the choice and give you the option to provide some initial feedback.

If it’s chosen something really odd, then you can also undo the categorisation and it will move it back to an unclassified state.

5

What taxonomy will be used during this program?

Your default taxonomy structure will be set up and configured for use with AI. If you have multiple, only the default will be configured at this stage.

6

Are there limits on what I can classify?

Yes, for now there are a couple you need to be aware of:

Firstly, only unclassified data will be processed. If a rule has already been created and is targeting spend, this will not be processed via the automated classification process.

Secondly, you will only be able to select up to 50 lines at a time.

Thirdly, there will be a weekly limit of 5,000 lines per week.

This is to ensure we keep costs controlled during this program whilst still gathering valuable feedback.

7

What level of detail does it classify at?

Line item level - the most granular way to classify data!

8

What level in my taxonomy will it try to classify at?

The feature will always try to find the lowest available level thats suitable even if you have 4 or 5 levels in your taxonomy structure.

9

What happens if the classification engine can’t find a suitable location?

It happens occasionally and the system will continue to show the data as unclassified. An icon will appear to show that the system has tried to classify the data but it hasn’t worked.

This is useful feedback for us as we can then investigate the reasons why it’s not worked.

10

Is my data secure?

Yes, we have deployed Azure Open AI within our existing Azure environment. Your data doesn’t leave the Rosslyn environment and is encrypted throughout. Your data is NOT being used by Open AI to train future models!

11

What happens if I don’t agree with the chosen categorisation?

There are two features available in this scenario:

1. The undo button - this will revert the classification and the line item will go back to an unclassified state
2. Provide feedback on the item - select the cross icon to let Rosslyn know you don’t agree with the chosen categorisation - this is very helpful for us to analyse and understand how the classification can be better in the future