Decipher training - which environment

[XavierGruchet]

Dear all,

We are currently implementing Decipher in 2 environments: test environment with test data and prod environment with prod data.

In order to create the different dfd, to do the testing of the different dfd etc, we are using the test environment. Meaning that we have to push prod data to a test environment which is theoretically not a good practice. We cannot use dummy data for this exercice as we need real pdfs. Once we are ready with the results, we push our configuration to prod and we can leverage the ML. We could even do the training in test and upload ML models to prod. Then we can just do slight modifications in prod.

We cannot do the testing of the different dfd in prod as it can mess up production processes and can have an impact on other functionnalities of BP that we are using.

Any thoughts what would be the best approach?  

Thanks
Xavier Gruchet

------------------------------
Xavier Gruchet
------------------------------

[lookman]

Hi Xavier.

It sounds like you have no choice but to use prod PDFs in your test environment. I think it is always better to NOT test in production if you can avoid it. So I say use prod PDFs in test to get your DFD dialed-in and then promote to prod along with the training data.

jack

------------------------------
Jack Look
Sr Product Consultant
Blue Prism
------------------------------