We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

------------------------------
Stephen Spence Fellow
Fellow, Architecture
Travelport
Slough
+447740638397
------------------------------

Dear Steve, 

Thanks for sharing your thoughts about the PoC. If I may ask ask a novice question around this. If a customer purchases the Watsonx assistant for IBM Z on z17. They need to train the model on z/OS or they get acess to a pre-trained model on the cloud. If the customer has cloud restrictid access, are they able to use to WA4Z ? 

Regards,
------------------------------
Abhishek Anand

Original Message:
Sent: Thu October 23, 2025 12:48 PM
From: Stephen Spence
Subject: POC of IBM Watsonx Assistant for Z (WA4Z)

We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

------------------------------
Stephen Spence Fellow
Fellow, Architecture
Travelport
Slough
+447740638397
------------------------------

We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

Thanks for sharing! Can you also talk a bit about what was actually deployed/installed on-premises?

We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

Hi Javier,

This was only a POC and not a production implementation which has a lot more capabilities e.g. Service Now integration.

We used IBM Cloud for the POC, this meant IBM quickly got this set  up and ready to support us for z/OS, z/TPF and z/VM. It also meant IBM experts in the WA4Z governance could change the weighting across the platforms and ingest our own documentation for us.

It's well work asking about a POC for your business!

Steve

------------------------------
Stephen Spence Fellow
Fellow, Architecture
Travelport
Slough
+447740638397
------------------------------

Original Message:
Sent: Tue October 28, 2025 11:26 AM
From: Javier Plumey
Subject: POC of IBM Watsonx Assistant for Z (WA4Z)

Thanks for sharing! Can you also talk a bit about what was actually deployed/installed on-premises?

------------------------------
Javier Plumey

Original Message:
Sent: Thu October 23, 2025 12:48 PM
From: Stephen Spence
Subject: POC of IBM Watsonx Assistant for Z (WA4Z)

We've completed a POC of the IBM Watsonx Assistant for Z (WA4Z) which used IBM Cloud for the AI.

First off, because we used IBM Cloud this was very fast to get started and the LLM was trained in z/TPF (my favourite operating system), z/VM and of course z/OS. This means it was immediately useable to us and after a few tries we had IBM use the governance tooling to alter the weighting in favour of z/TPF for us over z/OS.

At this stage novices in z/TPF were able to ask questions and get SME level explanations, I am a z/TPF SME and asked it some questions about things I did not know the answer to which generated good responses which included the source decument that was used as part of the response. I found the odd glitch in some answers which appeared to be due to the original document ingestion when building the LLM and these were addressed and corrected.

Next step was to import some company and 3rd party application documents into our instance of the LLM so we could then ask application specific questions too at which point we connected the POC to our applications tooling and encouraged developers to use it.

Overall this ws a very worthwhile POC and I thank our IBM account team for arranging it and the IBM WA4Z team for supporting the POC so well.

I'm happy to answer any questions on this!

Stephen #IBMChampion

------------------------------
Stephen Spence Fellow
Fellow, Architecture
Travelport
Slough
+447740638397
------------------------------