Custom copilot Teams Integration in Government applications
summary
Two government customers build and deploy Teams-based copilots using Azure OpenAI (GPT‑4 and GPT‑3.5), Azure AI Search/vector retrieval, semantic kernels and the Teams AI library; one prototype stands up in days using GitHub samples, app service, Key Vault, blob transcript logging and a Teams manifest. One deployment indexes about 5 million decisions for courts with narrow citations and Copilot plugin links; another reports productivity gains of about 2–5 hours per week per user and production-ready secure GCC High configuration.
AI-written summary · about
chapters (30)
- 0:00Custom Copilot Teams intro
- 1:18L3Harris AI assistant
- 3:44Azure OpenAI and Teams choice
- 4:47Security and productivity priorities
- 5:49Development journey and GitHub
- 6:54Bleeding edge AI adjustments
- 7:55Architecture deep dive and repos
- 11:10Government cloud security setup
- 12:39Core bot and Semantic Kernel
- 14:53MVP features and error handling
- 16:41Teams manifest and adoption
- 18:48Live demo: L3Harris Copilot
- 19:31Accelerating MVP with GitHub
- 20:13Early AI adoption advice
- 20:55Measuring AI ROI
- 21:17Future plans: internal data
- 22:21Studio 365 and Court of Justice intro
- 24:07Jurisprudence GPT overview
- 25:51Live demo: Jurisprudence GPT
- 27:21Azure OpenAI impact on dev
- 28:04Court of Justice reaction
- 28:46User feedback and search
- 29:07Productivity gains expected
- 30:11Learning curve and team
- 30:57Tech and architecture overview
- 31:42Development tools and process
- 32:03Copilot for M365 integration
- 33:08Future plans for Jurisprudence GPT
- 34:11Copilot dev advice
- 34:53Closing remarks
related talks
45mThe era of the AI Copilot
43mHarness the power of AI: Extend Copilot and beyond
45mExploring Copilot Capabilities
01h 51mShaping the future of work with AI
46mBuild and maintain your company Copilot with Azure ML and GPT-4
01h 01mUnlocking the Potential of AI
51mThe Best Way To Build An AI Copilot Product? Fusion Teams!
10mAI Co-Pilot Demo
