MPHS AI Power Day: Prompts → Prototypes → Productivity

On 25 Sept 2025, ~40 MPHS officers dived into a one-day, lab-first AI sprint tailored to council workflows, moving from smart prompting and source-aware research to Google Workspace AI and Canva outputs, then rapid prototyping with simple APIs and local models on Ollama—all wrapped with practical guardrails for safe rollout; everyone left with prompt packs, on-brand templates, starter code and a governance checklist to turn ideas into working results, fast.

9/25/20252 min read

Date: 25 September 2025
Location: MPHS, Selangor
Participants: ~40 officers

We just wrapped a focused, hands-on one-day AI training with Majlis Perbandaran Hulu Selangor (MPHS)—the same battle-tested programme we run for public-sector teams, tailored specifically for MPHS workflows so staff can turn ideas into working outputs fast. No fluff—just repeatable workflows you can use the very next day.

What we covered

  • GenAI toolbelt for daily work — Practical prompting and research with leading models and platforms (ChatGPT, Gemini, Claude, Perplexity, etc.) to draft memos, summarise long reads, compare sources, and structure briefs.

  • Google Workspace AI, end-to-end — Gmail, Docs, Sheets, Slides, Gemini, NotebookLM, and Google AI Studio: from inbox triage to circulars and simple data analysis—all inside tools MPHS already uses.

  • Canva AI for communications — Magic-assisted design (Magic Write, “Design for me”, image generation) to ship on-brand banners, social posts, and notices quickly.

  • Prototyping drills — API basics, a minimal feature calling an AI API, a tiny chatbot on Replit, local LLMs with Ollama, and lightweight orchestration with n8n/Flowise.

  • Safety, governance, and rollout — Prompt hygiene, data handling, red-teaming, evaluation, and a simple adoption plan that scales responsibly across units.

Why it matters for local-government teams

  • Speed with traceability — Faster turnarounds for briefs, responses, and talking points, with sources and attributions kept in view.

  • Consistency at scale — Shared prompts, templates, and checklists to standardise outputs across departments (licensing, enforcement, communications, ICT).

  • Right-sized innovation — Start with low-risk wins in existing tools; expand to prototypes once workflows are stable and governed.

In-room wins we loved

  • Turning a rough prompt into a well-structured memo/circular—with citations to verify every claim.

  • Converting briefing notes into a clean council-update slide deck in minutes.

  • Summarising heavy inbox threads (permohonan/aduan) into clear action items and next steps.

  • Spinning up a tiny chatbot and testing a local model, enough to demystify “how it all works”.

  • Using BM/English prompt patterns to keep outputs consistent across units.

What participants took home

  • Prompt packs & playbooks for day-to-day tasks (research, drafting, summarising, meeting notes, Q&A).

  • Templates for Docs/Slides/Sheets and design presets for quick public-facing comms.

  • Starter snippets for API calls and local-model experiments.

  • A governance checklist to keep outputs accurate, secure, and compliant.

Want this for your team?

We tailor one-day or two-day, lab-first programmes to your workflows—policy, comms, or IT. If you’re ready to move from slides to working results, let’s talk. We’ll bring the prompts, labs, and guardrails; your team brings the problems that matter.