FoundationaLLM Custom Copilot Workshop

This is the content for the 1 day workshop of FoundationaLLM Custom Copilot for Microsoft

Important

FoundationaLLM release cycle is every 2 weeks, so the content may change and be updated frequently until the first delivery on January 31st 2025 in New York City.

Note

This repository is private and only available to the Microsoft team and the FoundationaLLM team. It will be available to attendees during the workshops only.

Requirements

50 Entra IDs will be generated for the attendees to use during the workshop. The attendees will be able to use the Entra ID to login to the Management Portal and Chat Portal. A spreadheet will be shared with the attendees with the Entra ID and the password to register their name for one of the IDs.

Agenda

Lecture1: Where everything is, how to deploy (no deployment), Explain the resource Group and its resources and demonstrate the Management Portal, including Branding.

Lab1: Management Portal Discovery and capabilities - RBAC - Permission and Access (Share between 2 users to give access to each other)

Lecture2: Demonstrate the Chat Portal and its capabilities

Lab2: Chat Portal Discovery and capabilities

Note

Lab 1 and 2 will excersise 8 different excercises from generation of code to 10k analylics, to image recognition, to summarization, etc...

Lecture3: Explain the GatewayAPI, OrchestrationAPI and other APIs that make up the platform and demonstrate the creation of an Agent and every aspect of choices in the management Portal.

Lab3: Create an agent with OpenAI Assistant capability and use it in Chat Portal

Note

Lab 3 will include the ability of generating contract templates, using the template to generate a specific contract with details and finally download the word document version of the contract.

Lecture4: Go back to storage account and demonstrate the ability to add documents to a container and use POSTMAN to fire up the vectorization process while explaining the 4 stages and monitoring the progress in the vectorization-status

Lab4: Creating a Datasource and vectorization process

Note

Lab 4 will be about using data for Customer Service organization with RAG capabilities to vectorize content and use for answering questions.

Lecture5: Explain the Content safety approach and the efforts put in for Red Teaming the platform and also utilize Presideo for PII content protection.

Lab5: Content Safety and Presideo