Skip to main content

Chapter 39 — Productivity & The Agentic Office

Teaching Aid

"Every AI assistant I have tried knows everything about the world and nothing about my world. It can write a perfect OKR framework but doesn't know that we call our quarterly targets 'Boulders' not OKRs. It can produce a flawless meeting summary but doesn't know that when Zara says 'let's take this offline' she means it's politically sensitive and should not be minuted. It knows how to draft a project update but doesn't know that Project Nighthawk is what we call the Karachi expansion internally. I spend five minutes of every conversation re-explaining things that a good colleague would already know."

— Chief of Staff, 300-person technology company

Every domain chapter in Part 3 solved a specific professional problem. Chapter 28 deployed AI for financial analysis. Chapter 34 automated revenue operations. Chapter 35 optimised supply chains. Each is valuable in isolation. Together, they represent an AI-native organisation where every major work function has a layer of intelligent automation.

But there is a gap. Every one of those domain agents starts each conversation from zero. No memory of yesterday. No knowledge of your organisation's specific terminology. No awareness of project priorities.

This chapter closes that gap. It introduces the Workplace Memory Architecture — a four-layer context system that gives Claude persistent, structured knowledge of your people, projects, terminology, and priorities. It deploys 9 skills and 4 persistent agents that transform isolated domain tools into a coordinated Digital Chief of Staff — an AI that acts like a knowledgeable colleague, not a generic chatbot.

What You'll Learn

By the end of this chapter, you will be able to:

  • Diagnose the four context failure modes (terminology blindness, people anonymity, project amnesia, priority blindness) and explain why they persist despite AI capability
  • Deploy a two-plugin architecture — Anthropic's official Productivity plugin for task and memory infrastructure, plus the custom Agentic Office plugin for professional intelligence
  • Build the four-layer Workplace Memory (work.local.md) — personal, team, projects, and organisational context — and test it with stakeholder-aware outputs
  • Run task intelligence (brain dump capture with P1/P2/P3 priority sorting) and delegation workflows with handoff communications calibrated to each person's style
  • Generate daily digests, meeting intelligence (before/during/after), and executive dashboards with RAG status
  • Load cross-domain context that bridges Finance, HR, Operations, and Sales agents into integrated intelligence
  • Deploy four persistent agents — Chief of Staff, Memory Keeper, Meeting Intelligence, and Work Tracker — that maintain your workplace intelligence layer continuously

Lesson Flow

LessonTitleDurationWhat You'll Walk Away With
L01The Context Problem25 minThe four failure modes diagnosed; why domain agents alone are not enough
L02Two Plugins, One System20 minBoth plugins installed; TASKS.md, CLAUDE.md, memory/, dashboard.html, work.local.md all created
L03Workplace Memory Architecture40 minLayers 1 (Personal) and 4 (Organisational) built; terminology dictionary populated
L04Building Your People Memory35 minLayer 2 (Team) built; 5+ person entries with communication styles; person briefs tested
L05Projects and Priorities35 minLayer 3 (Projects) built; work.local.md foundation complete; cross-context search working
L06Task Intelligence40 minBrain dump captured and prioritised; P1/P2/P3 sort; delegation candidates identified
L07Delegation as a Discipline35 minDelegation records with calibrated handoff communications; follow-up protocol defined
L08The Daily Digest35 minDigest configured; first morning briefing generated; Monday/Friday variants understood
L09Meeting Intelligence40 minMeeting prep brief generated; D/A/F/Q/R coding used; structured synthesis produced
L10The Executive Dashboard40 minDashboard configured; RAG status for all projects; blocker classification understood
L11Cross-Domain Intelligence40 minCross-domain context loaded; integration gaps identified; search across all memory layers
L12The Digital Chief of Staff35 minChief of Staff agent configured; week-ahead brief and week-close summary generated
L13The Supporting Agents35 minMemory Keeper, Meeting Intelligence, Work Tracker configured; weekly maintenance cadence set
L14The Complete Agentic Office90 minFull integration smoke test passed; triggers, thresholds, and maintenance cadence defined
L15Summary and Quick Reference15 minAll commands, agents, memory layers, and the chapter's central insight

Chapter Contract

By the end of this chapter, you should be able to answer these five questions:

  1. What are the four context failure modes, and how does the Workplace Memory Architecture (four layers in work.local.md) address each one?
  2. How do the official Productivity plugin and the custom Agentic Office plugin divide responsibility, and why is zero trigger overlap between them critical?
  3. How does the delegation quality standard — a 7-item checklist with handoff communication calibrated to the delegatee's work.local.md profile — produce better outcomes than ad hoc delegation?
  4. What are the three phases of meeting intelligence (before/during/after), and how does the D/A/F/Q/R coding system ensure every meeting produces searchable, numbered decisions and owned actions?
  5. How do the four persistent agents (Chief of Staff, Memory Keeper, Meeting Intelligence, Work Tracker) work together to maintain continuous workplace intelligence rather than session-by-session briefings?

Prerequisites: Cowork Access

This chapter requires Cowork (set up in Chapter 28) and two plugins.

  1. Install the official Productivity plugin. In the Cowork sidebar: CustomizeBrowse plugins → find Productivity (from knowledge-work-plugins) → click Install. Alternatively: claude plugins add knowledge-work-plugins/productivity
  2. Install the Agentic Office plugin. In the Cowork sidebar: CustomizeBrowse pluginsPersonal → click +Add marketplace from GitHub → enter https://github.com/panaversity/agentfactory-business-plugins → find Agentic Office → click Install.
  3. Connect a working folder for practice files, same as Chapter 28.

Case Studies

Case StudyRolePurpose
Zia Khan (CEO, Panaversity / COO, PIAIC)Perspective characterDirect, evidence-based leader coordinating book production, campus expansion, and workshops
Omar Farooq (Head of Analytics)Key stakeholderData-driven, needs lead time, scope-conscious — delegation and communication calibration target
Ayesha Raza (Senior Data Analyst)New hire in onboardingFirst 30 days; fintech-to-edtech transition; timely feedback critical
Dr. Sana Mirza (Head of Curriculum)New senior hireAcademic precision, PHM framework ownership, relationship mediation needed

Agent Output Taxonomy

Errors are discovered progressively across lessons. By L14, you can diagnose all five:

Error TypeDiscoveredDiagnostic Question
Terminology BlindnessL01"Did the output use our internal terms or generic equivalents?"
People AnonymityL04"Did the output calibrate to this person's communication style?"
Priority ConfusionL06"Did the output treat a P1 task with appropriate urgency?"
Context LossL09"Did the synthesis reference context from prior meetings?"
Integration GapL11"Did the cross-domain context include all relevant domains?"

After Chapter 39

When you finish this chapter, your perspective shifts:

  1. You see the Context Problem everywhere. Every AI interaction that starts with "let me explain my situation" is a symptom of missing workplace memory. The question is no longer whether to build context — it is which layers to build first.
  2. You have a working two-plugin system. Task infrastructure (official) plus professional intelligence (custom) — 9 skills and 4 agents installed, configured, and producing outputs calibrated to your organisation.
  3. You understand the boundaries. The agents maintain memory, prepare meetings, track delegations, and assemble briefings. They do not make decisions, approve budgets, or manage performance. These boundaries are encoded in every skill and agent file.
  4. You can extend. The workplace memory architecture transfers to any context where persistent, structured knowledge improves AI output quality. The four-agent pattern applies wherever continuous monitoring creates value.

Start with Lesson 1: The Context Problem.