お問い合わせを送信いただきありがとうございます！当社のスタッフがすぐにご連絡いたします。
予約を送信いただきありがとうございます！当社のスタッフがすぐにご連絡いたします。
期間 14 時間
コース概要
- Day 1: Microsoft 365 Copilot and knowledge agents in Agent Builder
- Work cases and limits
- Clarifying licenses, approved data sources and permissions
- Selecting work cases, for example email, files, market analysis or document review
- What Copilot Chat, Agent Builder and Copilot Studio each offer
- Exercise: List the own work cases with source, desired result and control point.
- Copilot Chat and app Copilots
- The right interface for the task
- Email, meeting and task work
- Word, Excel and SharePoint as an information base
- Exercise: Solve a work case in Copilot Chat and note why an agent would add more value here.
- Agent Builder
- Audience, scope of tasks, instructions and answer format
- Knowledge sources such as SharePoint, files and web pages
- Conversation starters and limits
- Testing, sharing and permissions
- Exercise: Build a knowledge agent for process or market questions from released documents.
- Documenting the tool decision
- For each case: Copilot function, data source, result and control point
- What can be used directly and what needs a Studio agent
- Collecting open questions for IT and data protection
- Exercise: Create a tool map for the own work cases.
- Outcome: Each participant has a tested knowledge agent and a tool map for their own work cases.
- Work cases and limits
- Day 2: Copilot Studio, processes and control points
- Process analysis
- From document to reviewable process
- Mandatory fields, exceptions and open points in a review sheet
- The professional decision stays with the human
- Exercise: Break down a document process from the own area into steps with control points.
- Copilot Studio
- Instructions, topics, knowledge sources, tools and actions
- Branching and escalation when information is missing
- Publishing in Teams and Microsoft 365 Copilot
- Exercise: Design and build a document review agent in Copilot Studio.
- Agent flows and connectors
- Interplay of flow, connector and agent
- Human approval before every action in the target system
- Difference between information access and action
- Exercise: Extend the agent with a flow that includes approval via Teams or email.
- Test workshop
- Test cases with expected results
- Cross-checking against the source and failure cases
- A clear termination and handover point
- Exercise: Run through two test cases with anonymized data and record acceptance criteria.
- Operation and roadmap
- Roles, permissions, logging and billing
- Classification into now, later or deliberately not
- Carrying the knowledge into the team
- Exercise: Create a roadmap with three categories for the own use cases.
- Outcome: Each participant has a draft for a Studio agent with approval, two tested process cases and a roadmap for their own use cases.
- Process analysis
要求
- Microsoft 365 Copilot license, access to Copilot Studio (for day 2), released sample emails, files and documents, and a process from the own department.
- No programming skills required.
Audience
- Key users, AI champions and department leads who want to build their own Copilot agents for knowledge questions and document processes.
お客様の声 (1)
トレーナーは忍耐強く、とても親切です。彼はその分野に精通しています。
CLIFFORD TABARES - Universal Leaf Philippines, Inc.
コース - Agentic AI for Business Automation: Use Cases & Integration
機械翻訳
