Skip to content
PROGRAM 18 · TEAM TRAINING

AI & Data for Government & Public Service Officers

Clearer letters, faster reports, better service for citizens.

Important: Government data and classified documents must never go into public AI tools. The course follows the office's approved tools and data rules, and every official document is signed off by the responsible officer.

Who books it: Ministries, divisions, directorates, district and upazila offices, public projects and development partner programmes

8 sessions3 modules4 hands-on projects3 days · 9 am–5 pmCertificate on completion
CHOOSE A FORMAT
Request a proposalSee the curriculum

Who this course is for

  • Officers and staff in ministries, divisions and directorates
  • District and upazila administration and field offices
  • Project directors, PIU staff and monitoring officers in development projects
  • Public corporations, autonomous bodies and regulatory authorities

What your team will be able to do

  • Draft official letters, notes, minutes and circulars in clear Bangla and English
  • Turn long reports and policy documents into short briefs for decision makers
  • Analyse project, budget and service delivery data in Excel
  • Deploy four working automations within your office's approved tools
How it runs

3 days at your office, 8 sessions

Three sessions a day plus a hands-on build lab every afternoon. Day 3 ends with capstone presentations.

Day 1

9:00 am – 5:00 pm

AI chatbots for public service work

9:00–10:30S1Prompt engineering for public service
10:30Tea break
10:45–12:15S2Official letters, notes & minutes
12:15Lunch & prayer
1:00–2:30S3Citizen communication & service
2:45–4:30LABBuild lab: official letter and service guide on your files
4:30–5:00Q&A and day recap

Day 2

9:00 am – 5:00 pm

Data analysis, then the first two builds

9:00–10:30S4Formulas & public sector data cleaning
10:30Tea break
10:45–12:15S5Public sector analytics & decision notes
12:15Lunch & prayer
1:00–2:30S6Project 01 & 02: Rules assistant & alerts
2:45–4:30LABBuild lab: service dashboard, assistant and alerts
4:30–5:00Q&A and day recap

Day 3

9:00 am – 5:00 pm

Last two builds and the capstone

9:00–10:30S7Project 03: File & request approvals
10:30Tea break
10:45–12:15S8Project 04: Report & letter extraction
12:15Lunch & prayer
1:00–3:00LABCapstone build clinic
3:15–4:30LABCapstone presentations
4:30–5:00Certificates and next steps
Group: 10–40 people from one organisationSame content and projects in both formatsBangla delivery with English technical termsEvery session ends with a deliverable
Curriculum

8 sessions, one build in each

3 modules · 8 sessions · 4 hands-on projects

PART 1AI Chatbots for Public Service WorkSessions 1–3

Chatbot Landscape & Tool Selection

  • Gemini, ChatGPT, Claude, Copilot and DeepSeek: strengths for public service work (formal Bangla drafting, long report summaries, data tables, translation)
  • Free vs paid tiers; a workflow that needs no international payment card
  • Bangla vs English output quality; when to draft in English and translate
  • Which tool for a long document, which for a formula, which for a polished letter

Prompt Engineering Fundamentals

  • What a prompt is; why the same question gives different answers
  • The four-part structure: role, context, task, output format
  • Role prompting: act as a section officer, project director, monitoring officer or public information officer
  • Controlling output: length, tone, language, table or bullet format

Advanced Prompting Techniques

  • Few-shot prompting: teaching the model with examples of your own public service work
  • Step-by-step reasoning for multi-part tasks such as comparing physical and financial progress across project components
  • Constraint prompting: what to include, what to exclude, what to never assume
  • Chained prompts: breaking a progress report (data → findings → issues → recommendations → summary note) into a sequence of smaller steps

Iteration & Prompt Library Building

  • Why the second prompt matters more than the first; refining instead of restarting
  • Diagnosing a bad output: vague instruction, missing context or wrong tool
  • Building reusable prompt templates for repeated public service tasks
  • Testing a prompt before it goes into daily use

Data Privacy & Verification Discipline (Core)

  • What must never be pasted: classified or confidential files, citizen NID numbers and personal details, draft policy decisions, tender evaluations, staff personnel records
  • Anonymising data before prompting; safe vs unsafe public service use cases
  • Hallucination awareness; verifying every figure, date, name and legal reference
  • The verification rule: AI drafts, a person checks before anything leaves the building
DELIVERABLEa personal prompt library of 10 templates for your repeated public service tasks.
PART 2Public Service Data Analysis in Excel & Google SheetsSessions 4–5
PART 3Applied Projects: AI Assistants & AutomationSessions 6–8
Applied projects

Four automations, running in your organisation

01Session 6

Office Rules & Service Assistant

Officers, desk staff and field offices ask about service rules and get answers from office manuals and the citizen charter, with the section cited.

  1. INManuals, rules & citizen charter
  2. AIGrounded Gemini assistant
  3. OUTAnswer with the section cited
Built with Google AI Studio
02Session 6

Office Deadline & Follow-up Engine

Meeting decisions, ministry and IMED report dates, audit objection replies and RTI deadlines checked daily, with alerts 30, 15 and 7 days ahead.

  1. INDeadline & follow-up sheet
  2. RUNDaily Make scenario
  3. OUTAlert to the right owner
Built with Make · Sheets · Gmail
03Session 7

File & Request Approval Workflow

Internal requests routed to the desk officer, then the approving officer by type, with every step logged and pending files flagged.

  1. INInternal request form
  2. RULESApproval routing script
  3. OUTLogged decision in a register
Built with Apps Script · Forms
04Session 8

Report & Letter Extraction

Field reports and incoming letters become clean rows, with missing figures flagged and a follow-up entry for every letter needing action.

  1. INField reports & letters
  2. AIGemini structured extraction
  3. OUTChecked rows with follow-ups
Built with Gemini API · Drive

Capstone presentation

Each person presents one automation built for their own workplace.

Day 3 afternoon
Applied across every session

Responsible & Ethical AI in Public Service

Fairness & Non-Discrimination

  • Where bias enters public service decisions: data, wording, criteria and past patterns
  • Proxy discrimination: address, age, gender or background inferred indirectly

Privacy & Data Protection

  • Data minimisation: collect only what the decision requires
  • Retention and deletion of citizen, beneficiary and staff data

Public Trust, Equity & Transparency

  • Equal service for every citizen regardless of location, language or status
  • No automated decisions on benefits, licences or entitlements

Accountability & Human Oversight

  • Human-in-the-loop by design: AI recommends, a named person decides
  • Who owns an AI-assisted decision when it is challenged

Workforce Impact & Your AI Use Policy

  • Talking to public service teams about AI without triggering job-loss fear
  • Reskilling and role redesign as AI absorbs routine work
Outcomes

By the end, your team can

  • Choose the right AI tool and write prompts that work first time
  • Maintain a prompt library covering your repeated public service tasks
  • Draft official letters, notes, minutes and circulars in clear Bangla and English
  • Write citizen notices, service guides and FAQs people understand
  • Analyse project and service data and write a decision note
  • Build and deploy automations with Make, Apps Script and the Gemini API
  • Apply fairness, transparency and privacy to every AI-assisted decision

And leaves with

A template set of official letters, note sheets and minutes
A citizen service guide, FAQ and reply templates
A project or service dashboard with a decision note
A grounded office rules & service assistant your colleagues can use
A office deadline & follow-up engine that reaches the right owner
Four deployed automations running in your own organisation
A written AI use policy for your own office or project

Tools your team will use

Chatbots
ChatGPTGeminiClaudeCopilotDeepSeek
Content
CanvaGammaCapCutGoogle Slides
Data
Microsoft ExcelGoogle SheetsPower Query
Workspace
GmailCalendarMeetFormsDrive
Automation & AI builds
MakeGoogle AI StudioApps ScriptGemini API
Public Service tools
Bangla typing tools (AvroBijoy)Looker StudioGoogle Forms

What participants need

  • A laptop with Excel and a Google account
  • Basic Excel comfort and approval from your office to use the tools on non-confidential files
  • One real work file to practise on from Session 2
  • No coding background required; free tiers only

How your team will learn

  • build-along classes (you work while the trainer works)
  • practise on your own workplace file, not dummy data
  • every session ends with a deliverable
  • Bangla delivery with English technical terms
  • recordings, prompt library and setup guides
  • certificate after capstone submission
Request a proposal

Bring this program to your office

Tell us about your team. We reply within one working day with a plan built around your own documents and workflows.

  1. 1 · A short discovery call with your office lead
  2. 2 · We tailor examples to your masked files
  3. 3 · Offline at your office, or online live classes
Chat on WhatsApp
Request a proposal