Working notes · Azure & AWS

Grounding, agents and evaluation - with the numbers attached.

Notes from production deployments on Copilot Studio, Azure AI Search, Azure AI Foundry and Amazon Bedrock. Retrieval quality, agent routing, eval gates, and the invoices that follow - measured before claimed.

2
Posts: 4
Platforms: 2026
Since

2 postsCopilot Studio 0 postsAzure AI Search 0 postsAzure AI Foundry 0 postsAmazon Bedrock

LatestCopilot Studio·AUG 06 2026

Where to click in Application Insights for a Copilot Studio agent

Most of the Application Insights resource menu was built for SDK-instrumented web apps and is empty for Copilot Studio. Five blades are worth your time, every investigation ends in the same pane, and here is the route to it.

Read the piece →18 min · 3,602 words

All writing

2 posts · newest first

06 AugTroubleshooting a Copilot Studio agent before you reach for Application InsightsFour observability layers answer four different questions, and most defects are found in the first one. The habit that makes the rest tractable is capturing the conversation ID, because it is the only join key the four layers share.Copilot Studio·Deep dive·Agents, Governance35 min

Full archive →