How do you manage frequent sn_aia plugin updates and the high number of skipped records?

[GáborF]

Hi everyone,

I'm curios to hear how others are managing updates for the sn_aia plugins, especially considering how frequently new versions are released.

We work in two week sprints, with a change window at the end of each sprint. Our plugin update process follows this schedule, but we're finding it increasingly difficult to keep up with the release cycle.

By the time we complete the testing and deployment of one plugin version, a newer version is often already available. It feels like we're stuck in a never-ending update cycle. 

Why does the process take approximately two weeks?

Before reaching production, plugin updates go through tree non-production instance where we assess their impact, validate compatibility with our exsisting configuration, and make sure everything works as expected.
However, the most time consuming part of the process is reviewing the changes and the skipped records.

During a typical plugin update, we encounter anywhere between 800 and 2500 skipped records marked as Review needed. Going through these records and determining the appropriate action takes significant amount of time and effort.
This raises some concerns about the long-term sustainability of our current approach...
To make things more challanging, we frequently open support cases related to AI Agent functionality, and the most common recommendation we receive is to update the plugins to the latest version. While we understand the reasoning, keeping eveything up to date feels almost impossible given the effort required for each upgrade.

I'd really appreciate hearing how others approach this: 

• How frequently do you update you sn_aia plugins?
• Do you review every skipped records individually, or have you established a more efficient process?
• How do you handle the large number of skipped records marked as Review needed?
• Do you have any best practices for balancing plugin stability, testing requirements, and the need to  stay on the latest version?
• Has anyone found a sustainable way to manage this update cycle?

I'm particularly interested in hearing from teams managin AI Agent functionilty across multiple instances. 
Thanks in advance for sharing your experiences and recommendations!