Clean and Controlled AI with API Management
summary
Demo shows an Azure AI Gateway (API Management) fronting Microsoft Foundry and GPT-5 mini, enforcing a 60 tokens per minute per‑subscription limit (demo consumed 391 tokens and then blocked), blocking prompts via Content Safety blocklists in Cognitive Services, and semantic caching using a Redis enterprise cluster with a 0.6 similarity threshold and 60s cache duration. Configuration requires API Management Basic V2 (~€150/month), named values for Redis connection strings, and enables central auth, MCP server routing, and future model routing via AI Search vector indices.
AI-written summary · about
chapters (13)
- 0:00Intro and Speaker
- 0:38Nature Reserve AI Case
- 1:50Why Sustainable AI
- 4:13API Management Gateway
- 5:46Connect Foundry Endpoint
- 9:02Testing with Power App
- 9:59Token Rate Limiting
- 12:55Content Safety Blocklists
- 16:29Semantic Caching Setup
- 22:53Redis Tips and Costs
- 26:02Central Control and MCP
- 27:43Future Model Routing
- 29:34Wrap Up and Contact
related talks
16mAPI Management in the AI Era
31mAI APIs: Taming the Wild West with AI Gateways
43mUnleash the Potential of APIs with Azure API Management
01h 03mGoing production scale with Gen AI - Patterns for your API
30mAI APIs: Taming the Wild West with AI Gateways
31mAzure AI Innovations at Microsoft
33mPowering AI Capabilities with API Management
22mMaking AI Accountable: Azure AI Foundry