Agentic AI

This week we had a team meeting of the German special solution architects (like me) and the sales specialists. We met at a location that’s essentially a table tennis club retreat in Höhr Grenzhausen (near Montabaur). My manager knows this place because he’s a passionate table tennis player, and it’s perfect for a team event like this. The hotel was called Hotel Zugbrücke Grenzau, with a bar across the street where you could play bowling, billiards, or darts. There was also a large spa area and, of course, the table tennis hall. The food was excellent, and we also did a lot of sports activities.

There were many presentations, and one of them was mine, about Agentic AI, which I’d like to share here:

We generally split the topic between us - Manuel Hahn, a colleague, handled the demo section while I took care of the presentation. In our approach, we agreed to structure it the way I typically handle my presentations, starting with the end. So we first delved into the code in detail, allowing everyone to draw their own conclusions. There was a brief discussion, the resolution of which typically comes later, which experience has shown leads to better understanding and better retention of the topics.

This week also saw the Summit Connect, Red Hat’s conference in Boston, where several announcements were made. One of these announcements is related to the Agentic AI topic: Meta’s LLamastack. Red Hat and Meta have entered into a partnership in the AI space. Here’s the press release

As part of OpenShift AI, Red Hat will be releasing an operator that can manage the lifecycle Llama Stack distributions. Distributions are curations of well-tested providers for each of the LLama Stack APIs. Red Hat will offer/provide native support for a distribution, but will also allow for bring your own distributions for providers supported via the LLama Stack APIs, as well as partner distributions. Additionally, the operator will lifecycle server distributions per tenant.

All in all, this is a very good step in the right direction. I hope I was able to present this well in the presentation.