Category
Copilot, Copilot Studio, Azure AI Foundry and Azure AI Search.
4 articles
A working prompt in a playground is not a system. Production needs a way to tell whether a change helped, a deployment model that matches the traffic, content filtering you have configured deliberately, and observability into what is being spent.
Rajiv Kumar·Aug 5, 2026·10 min read
Copilot Studio makes it easy to build an agent that answers from documents. The gap between that and something people rely on is scope, grounded answers, real actions, and knowing what the agent must refuse to do.
Rajiv Kumar·Jul 8, 2026·11 min read
When a RAG system gives a wrong answer, the instinct is to change the prompt. Almost always the right passage was never retrieved. Chunking, hybrid search, filters and semantic ranking are the levers that actually move quality.
Rajiv Kumar·Jun 22, 2026·11 min read
Copilot searches what a user can already access. That sounds reassuring until you consider how much people can technically access and never find. Readiness work is mostly permission hygiene, and it is worth doing before the rollout, not after.
Rajiv Kumar·Jun 17, 2026·13 min read
