[Update Path] Operative NextGen

[aprildunnam]

[Update Path] Operative NextGen

Summary

Modernize the Agent Academy Operative learning path around a complete AI-assisted hiring system in the current Microsoft Copilot Studio experience. The path should progress from solution setup and agent instructions through skills, connected agents, responsible AI controls, resume intake, document generation, workflows, MCP scheduling, evaluation, monitoring, publishing, and completion.

This parent issue owns the path-level architecture, learning progression, shared standards, dependencies, and release readiness. Each mission specification is tracked as a child issue and remains independently authorable and reviewable.

Path outcome

A learner builds and validates a governed Hiring Agent and Interview Prep Agent that use Dataverse and Work IQ through MCP, process fictitious resumes, create hiring records, generate interview preparation material, automate intake, schedule meetings with confirmation, and demonstrate release quality through Evaluate and Monitor.

Audience

• Makers who completed Recruit NextGen or have equivalent Copilot Studio fundamentals.
• Agent builders ready for multi-agent orchestration, MCP, workflows, structured data, and evaluation.
• Teams that need a realistic but fictitious high-impact scenario for responsible agent development.

Ordered roadmap

Mission	Scope	Child issue
01	Get started with the Hiring Agent	#3151
02	Author agent instructions	#3148
03	Build skills and connected agents	#3149
04	Configure models, responses, and safety	#3150
05	Process resumes and applications	#3147
06	Generate an interview prep document	#3155
07	Automate hiring with workflows	#3152
08	Integrate with MCP servers	#3153
09	Obtain user feedback and evaluate	#3154
10	Publish, share, and complete the course	#3156

Path-level deliverables

• Ten authored and reviewed missions with one coherent hiring architecture.
• Reusable solution, Dataverse schema, sample records, fictitious resumes, and evaluation test sets.
• Current screenshots and UI labels for the validated tenant experience.
• Working cross-mission links, analytics tags, release checklist, and badge completion path.
• Accessibility review, Markdown validation, and a successful documentation build.
• End-to-end learner test covering the complete hiring scenario and cleanup.

Shared acceptance criteria

• Keep terminology, navigation, prerequisites, and screenshots aligned with the current Copilot Studio experience.
• Use the Hiring Agent as orchestrator and the Interview Prep Agent as the independently owned connected specialist.
• Explain when behavior belongs in agent instructions, a skill, a connected agent, an MCP tool, or a workflow.
• Use only fictitious candidate and employee data and exclude protected-characteristic inference.
• Require human confirmation immediately before consequential record creation, document release, or calendar booking where specified.
• Validate every product-specific step, tool name, permission, package, and preview dependency in a representative tenant.
• Include safety, privacy, failure, idempotency, and cleanup tests across the path.
• Preserve required mission metadata and analytics tags and maintain correct prerequisite and related-mission links.
• Run repository frontmatter and Markdown validation and npm run docs:build.
• Review the rendered path on desktop and mobile and complete one end-to-end learner walkthrough.

Dependencies

• Recruit NextGen completion or equivalent fundamentals.
• A Power Platform environment with Dataverse, solution import permissions, and seeded hiring data.
• Current Copilot Studio Skills, Connected Agents, Workflows, Evaluate, and Monitor experiences.
• Dataverse MCP and tenant availability of the Work IQ Calendar MCP server.
• Outlook, Teams, Microsoft 365 Copilot, and supported code execution where required by child missions.
• Product documentation and UI validation close to publication because labels and preview availability can change.

Out of scope

• Production applicant tracking system deployment.
• Autonomous hiring decisions or candidate rejection.
• Real candidate data or protected-characteristic processing.
• Organization-wide governance implementation, managed deployment pipelines, and production support commitments.

Risks

• Preview and tenant-specific capabilities may not be available consistently.
• Hiring data is sensitive and the scenario requires strict fictitious-data guidance.
• MCP tool names, permissions, and consent behavior can change dynamically.
• Python/code execution and document library support must be validated before Mission 06 is authored.
• Shared solution or Dataverse schema changes can cause cascading rework across missions.
• Workflow, calendar, and publishing tests create real side effects that require cleanup.

Definition of done

• All ten child issues are complete and linked to this parent.
• Shared solution assets, schemas, instructions, permissions, and test data are consistent across the path.
• Critical safety and end-to-end evaluation cases pass without unresolved regressions.
• No broken links, missing images, metadata errors, or build failures remain.
• A fresh learner account can complete the full path in the documented order.
• The path owner signs off on accuracy, accessibility, responsible AI, and release readiness.