Guides · 119
What are you trying to do?
Fix a problem
5The error or symptom in front of you, and the fix.AI & agents
Azure OpenAI 429 errors: why you're throttled below quota, and how to fix it
A 429 from Azure OpenAI rarely means you're out of quota. The rate limit counts estimated tokens including max_tokens, checks requests in one to ten second windows, and can be cut temporarily on shared capacity. Read the headers, then fix the real cause.
AI & agents
Azure OpenAI content filter errors: handle blocked prompts and cut-off answers
Azure OpenAI's content filter stops a request in two different ways: a 400 for a blocked prompt, and a 200 whose answer was cut. Handle both in code, read the annotations, know what the default guardrail does, and tune it in Foundry instead of fighting it.
AI & agents
Foundry Agent Service: build an agent, and move off the Assistants API
Build a prompt agent in Foundry Agent Service, then follow the migration path from the Azure OpenAI Assistants API: threads become conversations, runs become responses, a tool rewrites most of the code, and a few tools don't carry over.
AI & agents
Microsoft Foundry from zero: deploy a model, set quota and call it safely
Set up Microsoft Foundry from nothing: create the resource and a project, deploy a model with the right deployment type and quota, give your team the roles that actually let them build, and make the first call with Entra ID instead of a key.
AI & agents
RAG with Azure AI Search: give your Foundry agent your documents
Ground a Microsoft Foundry agent in your own documents with Azure AI Search: pick between the one-index search tool and a Foundry IQ knowledge base, build a vector index the agent can cite, connect it without keys, and fix the usual no-citations and 403 problems.
Quick reference
1Error codes, limits and cheat sheets to keep open.AI & agents
Microsoft Foundry for people who learned Azure OpenAI: what changed and what to migrate
A migration map from Azure OpenAI to Microsoft Foundry: every renamed thing side by side, the retirement dates that matter (the Assistants API, the azure-ai-inference package, stored completions), what to change in your code and resources, and what you can leave alone.