Post Content
In this interview we unpack what it really took to deliver our end‑to‑end “observe → evaluate → optimize” flow covering the end-to-end Agent DevOps lifecycle from inner‑loop offline signals to continuous improvement in production. We’ll share hard-earned lessons on (1) simplifying the getting started experience with out of the box observability powered by context-specific eval rubrics, (2) streamlining the developer experience with guided skill-based flows and (3) leveraging the complete set of inner and outer loop signals for continuous improvement.
𝗦𝗽𝗲𝗮𝗸𝗲𝗿𝘀:
* Vivek Bhadauria
* Filisha Shah
* Nitya Narasimhan
𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻:
This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com
LIVE159 | English (US)
Broadcast Stage
#MSBuild
Chapters:
0:00 – Why Agents Differ from Regular Software
00:02:44 – Using Evaluations to Monitor Agent Health
00:03:39 – Discussion on Continuous Optimization of Agents in Production
00:05:35 – Mention of BRK252 demo showcasing Agent Optimizer
00:05:49 – Agent Optimizer preview details – gated now, public by month-end
00:06:20 – Using user simulations and Trace Replace to identify harmful traces
00:06:47 – Discussion on favorite AI agent and learning through broken agents
00:07:33 – Conversation about developer collaboration and end-to-end observability
00:07:48 – Host emphasizes Build as an event for builders and asks colleague’s opinion Read More Microsoft Developer
