Corporate Onsite Syllabus

Effective Agentic Knowledge Work Workshop

2-Day Intensive Onsite 12–20 Knowledge Workers

This workshop teaches knowledge workers to wield general-purpose AI agents—Claude Cowork, Microsoft 365 Copilot / Copilot Cowork / Copilot Studio, ChatGPT Work, and Google Workspace Studio—as reliable, everyday problem-solving tools rather than chatbot toys. Participants will master the full knowledge-work agent lifecycle: task decomposition, context engineering, iterative delegation, output verification, and responsible rollout.

The curriculum is tool-agnostic by design. We teach principles that transfer across any agentic stack while providing hands-on experience with the leading platforms of 2026.

Engagement Model 2 × 2-hour online discovery sessions + 2-day onsite intensive

Duration 2 days (16 hours instruction + 4 hours guided lab)

Format Instructor-led, hands-on, cohort-based

Prerequisites None technical; participants should bring 2–3 real tasks they currently perform manually and want to automate or accelerate

Deliverables Personal agentic playbook, 2–3 working agent workflows, verification checklist, governance brief

Pre-Workshop Discovery Sessions (Online)

Session 1 — Scoping & Stakeholder Alignment

2 hours · 2–3 weeks before workshop

Audience mapping: Identify participant roles, skill levels, and daily workflows
Tool audit: Current AI tools in use (Copilot, Claude, ChatGPT, Google Workspace) and integration gaps
Task inventory: Collect 2–3 manual, repetitive, or time-consuming tasks from each participant
Pain point identification: Where does AI assistance currently break down or feel unreliable?
Platform constraints: Review approved tools, data-handling policies, and security guardrails

Deliverable: Customization brief and pre-workshop preparation checklist

Session 2 — Curriculum Customization & Environment Prep

2 hours · 1 week before workshop

Scenario finalization: Lock in 2–3 representative tasks to use as capstone candidates
Skill alignment: Identify which modules need depth vs. pace adjustment
Environment validation: Access to approved agent tools, sandbox accounts, sample data, and document repositories
Governance review: Confirm data policies, sharing rules, and approval workflows
Materials preview: Review custom playbooks, prompt templates, and evaluation rubrics built for your team

Deliverable: Finalized curriculum, customized materials, and environment readiness confirmation

Day 1 — The Agentic Mindset & The Knowledge Work Loop

Module 1: The Agentic Knowledge Work Landscape

90 min

Claude Cowork, Microsoft Copilot / Copilot Cowork / Copilot Studio, ChatGPT Work, Google Workspace Studio—and how to choose

From chatbot to coworker: why the shift to agentic execution matters
Tool comparison matrix: cloud vs. on-device, workspace integration, governance, cost, and data handling
Claude Cowork: autonomous, multi-step research, analysis, and content creation
Microsoft 365 Copilot / Copilot Cowork: agents inside Outlook, Teams, Excel, and SharePoint
Copilot Studio: building custom, low-code agents for specific business processes
ChatGPT Work / Workspace Agents: recurring workflow automation across Slack, Drive, and Notion
Google Workspace Studio: no-code flows across Gmail, Chat, Drive, and Sheets
The tool-selection decision tree: security, data residency, existing stack, team size

Hands-on: Compare the same task across 2–3 platforms; identify which tool fits which workflow

Module 2: Task Decomposition & Outcome Definition

90 min

The difference between a vague ask and a task an agent can reliably finish

The Knowledge Work Agent Loop: Define → Delegate → Verify → Iterate → Scale
Identifying agent-friendly tasks: repetition, clear inputs, verifiable outputs, bounded scope
Tasks to avoid (for now): ambiguous judgment, high-stakes decisions with no human check, sensitive data without governance
Writing outcome statements: what does "done" look like?
Breaking big tasks into verifiable sub-tasks
Defining success criteria and failure modes before delegating

Hands-on: Pick one manual task and reframe it as a delegable, verifiable outcome statement

Module 3: Context, Prompting & Iteration

120 min

How to give an agent the right information, in the right order, and steer it to consistency

Context engineering for knowledge work: background documents, examples, constraints, and audience
Prompt patterns for reliability: role, task, inputs, format, constraints, examples
The iterative refinement loop: first draft → targeted correction → verification → final polish
Working with documents, spreadsheets, and meeting transcripts as context
Managing long conversations: when to start fresh, when to thread, when to summarize
Using "show your work" and chain-of-thought techniques for transparency

Hands-on: Draft a multi-turn prompt sequence for a research, analysis, or writing task; compare output quality across iterations

Module 4: Building Reliable Single-Step & Multi-Step Agents

120 min

From one-off prompts to reusable, trustworthy workflows

Single-step agents: summarize, classify, extract, rewrite, compare
Multi-step agents: research → synthesize → draft → review → format
Building agents in common platforms: Copilot Studio, ChatGPT Workspace Agents, Google Workspace Studio, Claude Projects
Wiring inputs and outputs: documents, emails, forms, spreadsheets, chat channels
Adding human-in-the-loop checkpoints: approval gates, review steps, escalation triggers
Handling errors gracefully: when the agent stalls, hallucinates, or goes off-task

Hands-on: Build a multi-step agent/flow for one of the discovery-session tasks (e.g., meeting prep, report generation, email triage)

Day 2 — Applied Knowledge Work Agents & Responsible Scale

Module 5: Document, Data & Research Agents

120 min

Analysis, summarization, and reporting without the busywork

Document agents: summarize, compare, extract action items, draft from templates
Data agents: analyze spreadsheets, generate charts, explain variances, flag anomalies
Research agents: multi-source synthesis, competitive intelligence, policy scanning
Output formats: memo, slide deck, email, spreadsheet, dashboard-ready summary
Verifying agent-generated facts: source checking, citation tracing, and red flags
Working with proprietary data safely: what to upload, what to redact, what to avoid

Hands-on: Build a research or data-analysis agent that produces a structured report with verifiable sources

Module 6: Communication & Meeting Agents

90 min

Email triage, meeting prep, follow-up, and asynchronous updates

Email agents: prioritize, draft replies, flag escalations, extract commitments
Meeting agents: pre-meeting briefs, live notes, action-item extraction, follow-up emails
Chat agents: answer repetitive questions, route requests, surface knowledge-base answers
Tone and audience calibration: formal vs. casual, internal vs. external
Avoiding over-automation: when the human touch is non-negotiable

Hands-on: Configure a meeting-prep or follow-up workflow that runs from a calendar event or transcript

Module 7: Process & Cross-System Agents

120 min

HR, finance, operations, customer success, and other business workflows

Process automation: onboarding, approvals, status updates, routine reporting
Cross-system agents: connecting CRM, help desk, HRIS, and document stores via connectors and APIs
Using Model Context Protocol (MCP) and platform connectors to extend agent reach
Trigger-based vs. scheduled vs. on-demand agents
Measuring impact: time saved, error reduction, response-time improvement
Case-study deep dives: L'Oréal (conversational analytics), Thomson Reuters (legal research), Microsoft Copilot Studio customers

Hands-on: Map a cross-system process and build a trigger-based agent/flow using approved connectors

Module 8: Governance, Security & Scaling

90 min

Trust, auditability, and rollout discipline for agentic knowledge work

Data handling: what agents can and cannot see, sensitivity labels, retention policies
Human-in-the-loop governance: approval gates, mandatory review, escalation paths
Auditability: logging agent actions, version control for prompts and flows, traceability
Sharing and reuse: when to publish an agent, how to document it, how to prevent shadow AI
Change management: introducing agents without deskilling or alienating the team
Responsible use: bias, hallucination, over-reliance, and the "human still owns the decision" rule

Hands-on: Draft a governance brief for one agent/flow: data rules, review steps, audit log, and rollout plan

Optional Capstone Day (Day 3 — 6 hours)

Add-on package: Core + Capstone Day. Participants work in small teams or individually to take a real task from their role through the complete knowledge-work agent loop, guided by an instructor. Each participant/team must demonstrate:

Task definition: Clear outcome statement and success criteria
Tool selection: Rationale for choosing the platform(s) used
Agent build: A working multi-step agent/flow
Verification: Output checks, source validation, and human review steps
Governance plan: Data handling, approval gates, and audit approach
Rollout proposal: How to share, scale, and maintain the workflow

Instructor feedback: Real-time critique using the neurex.dev Agentic Knowledge Work Review framework

Post-Workshop Resources

Agentic knowledge work starter kit: task-decomposition worksheet, prompt templates, verification checklist
Platform configuration guides: Claude Cowork, Microsoft Copilot/Cowork/Studio, ChatGPT Work, Google Workspace Studio
Governance templates: data-handling policy brief, agent approval checklist, audit log template
Case-study library: enterprise examples across legal, finance, HR, marketing, and operations
30-day Slack access: Follow-up troubleshooting, prompt sharing, and pattern discussions