Classroom/ Online: Yes/ No
Scheduling Date(s):
Note: Please click specific date for detailed venue and course fee etc.
Claude for CPF Board
This workshop is designed to be delivered safely within CPF Board’s data and AI governance posture.
A Two-Day Hands-On Masterclass for Officers, Team Leads and Managers at the Central Provident Fund Board.
Day 1 builds the foundation — AI fundamentals applied to the daily operations of a Singapore statutory board.
Day 2 builds the automation layer — centred on Claude Cowork, with full attention to the data governance, PDPA, NAIS 2.0, AI Verify and IM8 considerations that bound CPF Board’s operating environment.
CPF Board officers carry one of the heaviest paper loads in the Singapore public service. Every policy change triggers FAQ updates, member communications, employer letters, internal SOPs, briefing notes, talking points, and contact-centre script revisions. Every channel — myCPF, contact centre, service centres, written correspondence — produces queries that require accurate, consistent, traceable responses. Every senior leadership engagement needs material drafted, refined, and updated as positions evolve.
This two-day masterclass teaches CPF Board officers and their leaders to apply Claude — Anthropic’s frontier AI assistant — to the daily work of running a Singapore statutory board.
Day 1 covers the fundamentals of AI applied to public sector operations: choosing the right model for the task at hand, configuring Claude with the officer’s working context (without exposing restricted data), writing prompts that produce work-grade drafts, and producing internal assets — briefing notes, draft replies, comparison tables, training material — that previously consumed an officer-day each.
Day 2 moves from operations to automation, centred on Claude Cowork — Claude’s desktop agentic surface that reads the officer’s own files in a folder, executes multi-step tasks in the background, and returns finished outputs to that folder. Each participant designs and runs at least one Cowork workflow against non-sensitive sample data representative of their unit’s work, and leaves with a 30-day adoption plan calibrated to CPF Board’s data classification regime and broader AI governance posture.
Public sector governance is built in, not bolted on. Every demonstration is set in a CPF Board–relevant scenario. Every exercise uses non-sensitive sample data. Every workflow taught is one that can be defensibly considered within the agency’s existing data and AI governance posture.
The three most common questions from prospective participants — answered directly:
A Two-Day Hands-On Masterclass for Officers, Team Leads and Managers at the Central Provident Fund Board.
Day 1 builds the foundation — AI fundamentals applied to the daily operations of a Singapore statutory board.
Day 2 builds the automation layer — centred on Claude Cowork, with full attention to the data governance, PDPA, NAIS 2.0, AI Verify and IM8 considerations that bound CPF Board’s operating environment.
CPF Board officers carry one of the heaviest paper loads in the Singapore public service. Every policy change triggers FAQ updates, member communications, employer letters, internal SOPs, briefing notes, talking points, and contact-centre script revisions. Every channel — myCPF, contact centre, service centres, written correspondence — produces queries that require accurate, consistent, traceable responses. Every senior leadership engagement needs material drafted, refined, and updated as positions evolve.
This two-day masterclass teaches CPF Board officers and their leaders to apply Claude — Anthropic’s frontier AI assistant — to the daily work of running a Singapore statutory board.
Day 1 covers the fundamentals of AI applied to public sector operations: choosing the right model for the task at hand, configuring Claude with the officer’s working context (without exposing restricted data), writing prompts that produce work-grade drafts, and producing internal assets — briefing notes, draft replies, comparison tables, training material — that previously consumed an officer-day each.
Day 2 moves from operations to automation, centred on Claude Cowork — Claude’s desktop agentic surface that reads the officer’s own files in a folder, executes multi-step tasks in the background, and returns finished outputs to that folder. Each participant designs and runs at least one Cowork workflow against non-sensitive sample data representative of their unit’s work, and leaves with a 30-day adoption plan calibrated to CPF Board’s data classification regime and broader AI governance posture.
Public sector governance is built in, not bolted on. Every demonstration is set in a CPF Board–relevant scenario. Every exercise uses non-sensitive sample data. Every workflow taught is one that can be defensibly considered within the agency’s existing data and AI governance posture.
The three most common questions from prospective participants — answered directly:
| Question | Where it is answered |
|---|---|
| Does this cover the fundamentals of AI and how AI can support daily operations at a Singapore statutory board? | Day 1 (M1–M6) covers the foundation and applies it throughout to CPF Board–relevant operations — FAQ drafting, briefing notes, member-reply templates, internal SOP updates, training material, and feedback summarisation. Day 2 (M7–M13) covers automation end to end. |
| What AI tools will the trainer use, and how do they align with PDPA, NAIS 2.0, AI Verify and IM8? | Section 7 lists every tool the trainer demonstrates, with the modules in which it is used. Section 8 sets out the operating environment, governance posture and data-handling rules observed throughout the workshop — no restricted or member-identifiable CPF data is used. |
| Does this course use Claude Cowork? | Yes. Cowork is the centrepiece of Day 2 — two live demos, two guided exercises, and one workflow each participant designs and runs against non-sensitive sample data representative of their unit’s work. |
Objective
By close of Day 2, participants will be able to:
- Apply the fundamentals of AI — capabilities, limits, and the AI Adoption Pyramid — to the officer’s daily operations at CPF Board.
- Select the right Claude model — Haiku, Sonnet, Opus or Adaptive — for the task at hand, with due regard to data sensitivity, latency and cost.
- Configure Claude with the officer’s working context — Memory, Voice, Projects — without exposing any classified, restricted or member-identifiable information.
- Use the 6P Prompt Framework and the 5-Step AI Thinking Partner Framework to produce work-grade drafts of briefing notes, FAQ updates, employer letters, talking points and internal SOPs.
- Use Artifacts, Skills and Claude Design to produce internal-facing assets — comparison tables, decision trees, training visuals — without external design support.
- Design and run Claude Cowork workflows against non-sensitive sample data representative of the officer’s unit work — including document production, data summarisation, multi-step research and SOP versioning.
- Apply Claude’s other agentic surfaces — Scheduled Tasks, Dispatch, Claude for Chrome and Claude Code — to extend automation responsibly, within the bounds of CPF
- Build a personal 30-day Claude adoption plan calibrated to the officer’s unit and CPF Board’s AI governance posture (PDPA, NAIS 2.0, AI Verify, IM8).
Outline
Day 1 — Foundations & Daily Operations at CPF Board
Day 2 — Operations & Automations with Claude Cowork
| # | Module | What’s Inside | Min |
|---|---|---|---|
| M1 | AI Fundamentals | How LLMs actually work, what they can and cannot do, the AI Adoption Pyramid, the 5-type learning taxonomy — framed for public sector officers. | 45 |
| M2 | Models & Memory | Haiku, Sonnet, Opus, Adaptive — when to use which, with explicit attention to data sensitivity, latency and cost trade-offs. Memory and Voice configured to the officer’s role. | 45 |
| M3 | Prompt Engineering | The 6P Prompt Framework and the 5-Step AI Thinking Partner Framework, applied live to CPF Board officer tasks — FAQ drafting, briefing notes, employer letters, member-reply templates. | 60 |
| M4 | Connect | Where Connectors fit in a statutory board setting — Microsoft 365 mail, calendar, files and the agency’s productivity stack — and the governance posture to apply. | 45 |
| M5 | Build | Projects, Artifacts and Live Visuals — produce internal-facing assets in the room: comparison tables of policy versions, decision trees for officer guidance, training visuals. | 60 |
| M6 | Customise | Skills and Claude Design (claude.ai/design) — turn one-off prompts into reusable, audit-ready capabilities the unit can share. | 45 |
Day 2 — Operations & Automations with Claude Cowork
| # | Module | What’s Inside | Min |
|---|---|---|---|
| M7 | Day 1 Recap & Cowork Primer | What Cowork is, the desktop folder model, safety and permissions — and how Cowork maps onto CPF Board’s data classification regime. | 45 |
| M8 | Cowork Hands-On #1 | Live demo + guided exercise: consolidating a folder of synthesised internal memos and circulars into a briefing note for senior management. | 60 |
| M9 | Cowork Hands-On #2 | Live demo + guided exercise: synthesising a folder of (synthesised) member-feedback or contact-centre data into themes, trends and recommended next actions. | 60 |
| M10 | Build Your Own Cowork Workflow | Each participant designs, builds and runs one Cowork workflow against non-sensitive sample data representative of their unit’s recurring work. | 75 |
| M11 | Browse & Automate | Claude for Chrome for officer web research; Scheduled Tasks for recurring outputs; Dispatch for mobile pairing — each with a governance note. | 60 |
| M12 | Extend | Claude Code for lightweight internal tools relevant to analytics and transformation teams; an honest discussion of where the agentic surfaces fit (and where they do not) inside a statutory board. | 45 | M13 | Adopt | Personal 30-day plan calibrated to the officer’s unit and CPF Board’s AI governance posture; Q&A and close. | 30 |
Who should attend
Recommended group size: 10 to 16 participants.
Recommended audiences: officers, team leads and managers at CPF Board across functions — including policy and operations, member services, employer services, contact-centre operations, communications and corporate affairs, learning & development, data and analytics, transformation and innovation, and IT/digital teams.
Best delivered as a closed in-house masterclass for a single CPF Board division or department, or as a cross-functional cohort drawing from multiple units. Equally suitable for leadership or officer-grade cohorts; mixed cohorts work well when framed unit-by-unit.
Recommended audiences: officers, team leads and managers at CPF Board across functions — including policy and operations, member services, employer services, contact-centre operations, communications and corporate affairs, learning & development, data and analytics, transformation and innovation, and IT/digital teams.
Best delivered as a closed in-house masterclass for a single CPF Board division or department, or as a cross-functional cohort drawing from multiple units. Equally suitable for leadership or officer-grade cohorts; mixed cohorts work well when framed unit-by-unit.
Methodology
The workshop runs three complementary modes of learning across both days.
Day 2 leans more heavily on practical sessions, given the automation focus. Each participant leaves Day 2 with at least one Cowork workflow they have personally built, tested and run.
Data posture: All exercises use non-sensitive sample data. No restricted, confidential or member-identifiable CPF data is processed during the workshop. Officers who wish to apply a method to their actual work do so afterwards, within their unit’s data governance posture and with the appropriate clearances.
AI Tools Used by the Trainer
Every tool below is demonstrated live during the workshop. Participants use the same tools on their own workshop laptops where applicable.
Scope note: No other AI vendors’ tools are used in the core workshop. Day 1 includes brief, neutral comparison points to other frontier models at the conceptual level (when each is genuinely better suited), but the workshop is delivered end to end on Claude. Consumer messaging channels (iMessage, Telegram, Discord) are mentioned only for completeness of the Claude product surface — they are not part of the practical content for CPF Board officers and are not used in any exercise.
Prerequisites
Participants should arrive with the following in place:
Lesson Plan
Day 1 — Foundations & Daily Operations
Day 2 — Operations & Automations with Claude Cowork
- Presentation — The facilitator demonstrates each capability live, ten to fifteen minutes per capability.
- Discussion — Facilitated debrief and peer exchange after every demonstration — with explicit attention to where each capability does and does not fit at CPF Board.
- Practical Sessions — Hands-on application on non-sensitive sample work representative of the officer’s unit. Every module includes a practical sprint; the facilitator coaches the room.
Day 2 leans more heavily on practical sessions, given the automation focus. Each participant leaves Day 2 with at least one Cowork workflow they have personally built, tested and run.
Data posture: All exercises use non-sensitive sample data. No restricted, confidential or member-identifiable CPF data is processed during the workshop. Officers who wish to apply a method to their actual work do so afterwards, within their unit’s data governance posture and with the appropriate clearances.
AI Tools Used by the Trainer
Every tool below is demonstrated live during the workshop. Participants use the same tools on their own workshop laptops where applicable.
| Tool | Used In | What it does in the workshop |
|---|---|---|
| Claude.ai (Pro / Max) | M2–M6 | Core chat surface — prompting, Projects, Artifacts, Skills, Design. |
| Claude Desktop App | M7–M11 | Hosts Cowork and Dispatch — required for Day 2 (subject to CPF Board IT approval; see Section 8). |
| Claude Cowork | M7–M10 | Agentic desktop assistant working against the officer’s own (non-sensitive sample) files — the centrepiece of Day 2. |
| Claude for Chrome | M11 | Agentic browser — officer web research, form-fill and multi-tab tasks. |
| Claude Code | M12 | Lightweight internal tools and automations — framed for analytics, data and transformation teams; production use is subject to IT engagement. |
| Scheduled Tasks | M11 | Set Claude to run recurring tasks (e.g. weekly summaries from a sample folder) automatically. |
| Dispatch | M11 | Mobile pairing — trigger Claude tasks from a smartphone (demonstration only; deployment is an IT matter). |
| Connectors | M4 | Microsoft 365 mail, calendar and files — demonstrated against a non-sensitive sandbox account; the production posture is a CPF Board IT decision. |
Scope note: No other AI vendors’ tools are used in the core workshop. Day 1 includes brief, neutral comparison points to other frontier models at the conceptual level (when each is genuinely better suited), but the workshop is delivered end to end on Claude. Consumer messaging channels (iMessage, Telegram, Discord) are mentioned only for completeness of the Claude product surface — they are not part of the practical content for CPF Board officers and are not used in any exercise.
Prerequisites
Participants should arrive with the following in place:
- Workshop-appropriate Claude.ai Pro or Max access for the duration of the workshop — provisioned by CPF Board L&D and IT
- A laptop suitable for the workshop, with Chrome and the Claude Desktop App installed in advance — in the configuration agreed in Section 8.
- Microsoft 365 mail and calendar access (sandbox or workshop account) for the Connect module.
- A smartphone for the Dispatch pairing demonstration (optional).
- Three to five non-sensitive sample documents representative of the participant’s unit work — e.g. publicly available CPF Board materials, sanitised templates, or items provided by the trainer.
- A working folder on the workshop device (e.g. ~/CPF-Workshop/) containing three to five non-sensitive sample files that Cowork will be given access to on Day 2.
- Basic comfort with any AI chat tool. No coding background required.
Lesson Plan
Day 1 — Foundations & Daily Operations
| Time | Block |
|---|---|
| 9:00 | Opening — pair intros: one recurring task the officer wants off their plate by end of Day 2 |
| 9:30 | M1 Fundamentals — How LLMs work, AI Adoption Pyramid, 5-type learning taxonomy |
| 10:15 | M2 Models & Memory — Haiku, Sonnet, Opus, Adaptive; Memory and Voice |
| 11:00 | Refreshment break |
| 11:15 | M3 Prompt Engineering — 6P Framework, 5-Step AI Thinking Partner, applied to CPF officer tasks |
| 12:15 | M4 Connect — Microsoft 365 and the agency productivity stack |
| 13:00 | Lunch |
| 14:00 | M5 Build — Projects, Artifacts, Live Visuals |
| 15:00 | Refreshment break |
| 15:15 | M6 Customise — Skills, Claude Design |
| 16:00 | Day 1 wrap, reflections, prep for Day 2 |
| 17:00 | End Day 1 |
Day 2 — Operations & Automations with Claude Cowork
| Time | Block |
|---|---|
| 9:00 | Opening recap — what stuck, what surprised, what to apply today |
| 9:15 | M7 Cowork Primer — what it is, the folder model, safety, and how it maps to CPF’s data classification regime |
| 10:00 | M8 Cowork Hands-On #1 — briefing-note production from a folder of synthesised internal memos |
| 11:00 | Refreshment break |
| 11:15 | M9 Cowork Hands-On #2 — synthesising synthesised member-feedback into themes and recommended actions |
| 12:15 | M10 (start) Build Your Own Cowork Workflow — design phase |
| 13:00 | Lunch |
| 14:00 | M10 (continue) — build, test, run |
| 14:30 | M11 Browse & Automate — Claude for Chrome, Scheduled Tasks, Dispatch (with governance notes) |
| 15:30 | Refreshment break |
| 15:45 | M12 Extend — Claude Code; honest fit-and-limits discussion of the agentic surfaces at CPF Board |
| 16:30 | M13 Adopt — Personal 30-day Plan, Q&A and Close |
| 17:00 | End Day 2 |
Profile of Andrew Chow
Andrew Chow CSP, MBA — Founder & CEO of AP Academy Pte Ltd, angel investor through Asia Pro Ventures Pte Ltd, Past President of Asia Professional Speakers Singapore, and author of Personal Branding 247.
Thirty-five-plus years building, training and advising organisations across ASEAN. Over 100,000 professionals trained in more than 20 countries.
Public sector experience includes engagements with Singapore statutory boards (including Home Team Academy) and an ongoing AI capability engagement with the Culture Academy at the Ministry of Culture, Community and Youth (MCCY). Andrew’s public-sector practice emphasises AI capability that fits within Singapore’s governance frameworks — PDPA, NAIS 2.0, AI Verify and IM8 — rather than capability that has to be retrofitted to them.
