Case study · 9 min read

Google Workspace AI Training for 30 People Across 9 Departments: Meenakshi Polymers

How a full-day, on-site Google Workspace and Gemini working session on 19 September 2026 gave Meenakshi Polymers' HR, Quality, Production, Logistics, Accounts, Operations, Legal, MIS and IT teams 27 hands-on exercises. With slides and photos from the day.

By Surendra Pandar, AI Trainer ·

Participants at the Meenakshi Polymers Google Workspace AI training session
Organisation
Meenakshi Polymers Pvt. Ltd.
Audience
About 30 people across 9 departments
Departments
HR, Quality, Production, Logistics, Accounts, Operations, Legal Compliance, MIS, IT
Date
19 September 2026, 9:30 AM to 5:00 PM
Format
On-site, full-day working session, 11 modules
Topic
Gemini, Gmail, Docs, Sheets, NotebookLM, Slides, Forms, Meet, Drive and Workspace Studio

In short

  • A full day, not a talk. The deck has 79 slides in 11 modules, and 27 of those slides are hands-on exercises that people did on their own laptops, in their own Google Workspace.
  • The day was built around one framework, not a list of tools: Role, Task, Context, Format. Every app after Gemini was taught as the same skill in a different place.
  • Each department worked on its own material: HR on policies and circulars, Quality on SOPs, Production on rejection data, Accounts and Legal on policies and contracts, IT on the questions it gets every week.
  • The day ended with four red lines and a list of the skills AI cannot replace, so the team left with confidence and with limits.

The challenge

Meenakshi Polymers already works inside Google Workspace every day: Gmail, Docs, Sheets and Meet. The AI inside those apps had arrived, but most of it was switched off or unused. The session opened with exactly that line: "You already pay for these tools. Most of it is switched off."

The room was not one team. About 30 people from nine departments sat together, from HR and Quality to Production, Logistics, Accounts, Operations, Legal Compliance, MIS and IT. A generic demo would have helped nobody. The brief was a working day where every person tries the tools on their own work and leaves with something that keeps working after the trainer has gone.

Slide: two things today. A framework, not a list of tools. AI becomes your friend, not a stranger.
Slide 2 set the two goals of the day: learn a framework rather than a list of tools, and make AI a friend rather than a stranger.
Introduction slide: who is talking to you, background, what I do and why I am here
The 60-second introduction, which ends on the reason for the day: most of what you already pay for is switched off.

How the day was built

The day ran from 9:30 AM to 5:00 PM with two tea breaks and a lunch break. The day plan was built so that about a quarter of the time was talking, a third was live demonstration, and about two-fifths was people doing the work themselves. The deck is organised in 11 modules, and every module after the first two has at least one exercise slide that starts with "Hands on".

ModuleWhat the room did
01 Why AI nowSeventy years of AI in one slide, predictive versus generative AI
02 How to talk to GeminiRole, Task, Context, Format; the same question asked as a lazy prompt and a full prompt
03 Making it permanentGems, scheduled actions, skills and Spark
04 GmailAsk the inbox, draft and refine, turn a mail into a calendar event
05 Google DocsFix a real SOP, write in your own voice
06 Google SheetsPlain-English formulas, questions to your data, and checking one answer yourself
07 NotebookLMCited answers from your own documents, Audio Overview
08 Slides and FormsA deck and a form built from one description
09 Calendar, Meet and DriveMeeting notes that write themselves, finding a file by describing it
10 Workspace StudioAn agent that starts on its own, with a person approving
11 Limits, and youWhere AI fails, four red lines, the skills only people have

Module 2: one framework for every app

The only framework in the day was Role, Task, Context, Format. It was taught once, in Gemini, and then used in every app that followed. The reasoning in the deck is simple: once people can talk to the AI, every AI button in every app starts to make sense within half a minute.

The first exercise was deliberately personal. Each participant picked one task from their own desk, wrote it using all four parts, and ran it. After that the room followed the follow-up habit: shorter, more formal, as a table, in Hindi.

Hands-on slide: write your first real prompt with the four parts Role, Task, Context and Format
Hands on: pick one task from your own desk and write it with all four parts.
Hands-on slide: upload a document, ask three questions, ask for the answer as a table
Hands on: upload any document from your laptop, ask three questions, then ask for a table.
Hands-on slide: run a Deep Research on a vendor, raw material, competitor or standard
Hands on: start a Deep Research on a real question from work and let it run while the session continues.

Module 3: from a demo to a habit

A demo impresses people for an afternoon. A saved Gem works for months. This module was where the day moved from trying to keeping. Participants picked the one thing they write again and again, wrote the four parts into a new Gem, ran it twice with two different inputs, and then fixed the instruction, not the prompt.

The point made on the slide: ten people using one Gem produce one format; ten people prompting fresh produce ten. For a company where reports go between departments, that is the whole business case.

Hands-on slide: build your own Gem in four steps
Hands on: build a Gem for something you write again and again. "You are keeping this one. It works after I leave."
Hands-on slide: hand one read-only task to Spark and let a person approve
Spark was handled carefully: check availability first, give it a read-only task, and let a person approve anything that leaves.

Gmail and Docs: the work people already do all day

Gmail was taught in three ways, as on the deck: search and pull information from your own mail and Drive, draft and write, and trigger the next thing. The exercise that people remember is "the thread you gave up on": find a long thread you never finished reading, summarise it, draft the mail you have been putting off, refine it once, then shorten it.

In Docs the room fixed a real document. An SOP from the shared folder was given headings, summarised in five bullets and turned into a responsibility table. The second exercise is the one HR and Sales said they would use every week: paste two documents your team actually wrote, ask Gemini to write the next one in the same voice, and compare it with your first draft.

Hands-on slide: ask your own inbox using the Gemini side panel in Gmail
Hands on in Gmail: ask a question about something buried in your own mail, and find a file whose name you forgot.
Hands-on slide: the thread you gave up on, summarise it and draft the reply you have been putting off
Hands on: the thread you gave up on.
Hands-on slide: open the SOP in the shared folder, apply headings, ask for a summary and a responsibility table
Hands on in Docs: fix a real SOP so it has structure, a summary and a table of responsibilities.

Gmail, the thread you gave up on

Summarise this thread in five bullets. List every open question and who it is waiting on. Then draft a reply from me that answers the first open question in under 100 words, in a polite and direct tone.

Our wording of the exercise on the slide, using the same Role, Task, Context, Format parts.

Sheets: the formula problem is over, the judgment problem is not

In Sheets the room used a production data file from the shared folder. The first exercise was "Stop remembering formulas": ask in plain English for a rejection percent formula and apply it, then ask for a count of one line in one month. The second was "Ask your data a question": which reject reason is the biggest and on which line, which shift is getting worse month on month, and a chart of rejection by shift.

Then came the slide that mattered most for a Quality team: check one answer yourself. Go into the data and verify what the AI told you. Sheets is where AI is weakest: excellent at the formula, not at the judgment. The number stays yours, especially in an audit.

Hands-on slide: open the production data, ask for a rejection percent formula, ask for a count of one line in one month
Hands on in Sheets: describe the formula in English and let it write the COUNTIFS.
Hands-on slide: which reject reason is the biggest and on which line, which shift is getting worse, chart of rejection by shift
Hands on: ask your data a question, then make a chart from a sentence.

NotebookLM, Slides, Forms, Meet and Drive

NotebookLM was positioned as the answer to everything the room was worried about at 10 AM. Gemini knows the world; NotebookLM knows only the documents you give it, and shows the page each answer came from. For a company full of SOPs, policies and audit files, that is the tool people took home. The exercise: generate an Audio Overview from your notebook, play 30 seconds, and ask yourself who in your team would actually listen to it.

Forms and Meet gave the quickest wins. In Forms, people typed one line describing the form they keep postponing, such as a safety observation, gate entry or leave request, watched the questions build, and sent it to the person next to them to fill. In Meet, two people started a 60-second call, turned on Take notes for me, talked about their biggest time waster, ended the call and found the notes in Drive. In Drive, they described a file instead of naming it.

Hands-on slide: turn a policy into a podcast with NotebookLM Audio Overview
Hands on in NotebookLM: turn a policy into an Audio Overview.
Hands-on slide: build the form you keep postponing from one line
Hands on in Forms: build the form you keep postponing, from one line.
Hands-on slide: start a 60-second call, turn on Take notes for me, find the notes in Drive
Hands on in Meet: start a 60-second call and let the meeting write itself.
Hands-on slide: use the Gemini panel in Drive to find a file by describing it
Hands on in Drive: describe a file instead of naming it.

Workspace Studio: the agent drafts, a person approves

Everything until this point needed a person to start it. The Workspace Studio module showed an agent that does not. The worked example on the slides is a complaint: a mail arrives from a customer domain, the agent pulls out the part number and the issue, posts it for Quality and drafts an acknowledgement.

The most important slide in the module was the one that says what the agent does not do: the agent drafts and a person approves. The advice to the room was to build every agent that way for the first six months, and to automate the typing, not the decision.

Slide: the agent drafts, a person approves
The rule for every agent in the room: the agent drafts, a person approves.

How different departments used the day

Because the exercises asked people to pick their own task, each department ended up on its own material. The table below shows how the day mapped to the nine departments in the room.

DepartmentWhere the day fitted their work
HRWriting in your own voice in Docs, policy questions in NotebookLM, induction questions
QualitySOP structure in Docs, checking one answer in Sheets, cited answers from procedures
ProductionRejection questions in Sheets, work instructions in plain language
LogisticsFinding files in Drive, packing and dispatch documents in NotebookLM
AccountsPlain-English formulas, policies and payment terms in NotebookLM
OperationsMeeting notes from Meet, open actions from long threads
Legal ComplianceContracts and checklists in NotebookLM, with the red lines on checking
MISCharts from a sentence, metric definitions
ITThe questions it gets every week, saved as a Gem or a form

This table is our summary of how the exercises map to each function. It describes the shape of the day and is not a measured result.

Limits and the close

The last module was taught with laptops closed. It opened with the line that AI cannot be held responsible and has no intent of its own, so a person stays accountable for anything that leaves the company. It listed four red lines, including reading everything before you send it because your name is on it, and a slide of skills AI cannot replace, starting with taste.

The closing slide brought back the opening idea: a friend is useful, but you still do not let a friend sign your cheques.

Slide: four red lines for using AI at work
Four red lines, the last module of the day.
Closing slide: a friend is useful, you still do not let a friend sign your cheques
The closing line of the day.

The result

Each person left with something that works without the trainer: a Gem for a task they repeat, a notebook of their own documents, and a list of tasks to try in Gmail, Docs, Sheets, Forms and Meet. Because the session used tools Meenakshi Polymers already had, there was no new software to buy. The saving comes from using existing subscriptions properly.

Sandeep Sareen, VP at Meenakshi Polymers, said the session gave a practical understanding of how Gemini can be used across Google Workspace, and that the automation examples made it easier to see where AI saves time. We have not published hours or money saved because they were not measured in the session.

Participants at the Meenakshi Polymers Google Workspace AI training session using Gemini on their laptops
Participants working with Gemini in Google Workspace during the session.

Takeaway for other organisations

If your company already uses Google Workspace or Microsoft 365, start there. Training people to use the AI inside tools they already own is often the lowest-cost way to save time with AI. Make the day a working day: every person should open their own mail, their own SOP and their own data, and finish with one thing saved that keeps working.

Want a similar day for your team? Read the full guide to Google Workspace and Gemini for business teams, then get in touch at /contact to plan a session for your departments.

“The session gave us a practical understanding of how Gemini can be used across Google Workspace. The automation examples made it easier to see where AI saves time.”

Sandeep Sareen, VP, Meenakshi Polymers

Frequently asked questions

How many people attended the Meenakshi Polymers training?

About 30 people from nine departments: HR, Quality, Production, Logistics, Accounts, Operations, Legal Compliance, MIS and IT.

Was the session online or on-site?

On-site, as a full working day on 19 September 2026, from 9:30 AM to 5:00 PM.

Do we need to buy new software to use AI at work?

Often not. Many teams already have AI features inside Google Workspace or Microsoft 365. Training is what turns those features into saved time.

Does every Workspace plan include every feature shown?

No. Some features, such as Spark and Workspace Studio, are rolled out in stages and depend on the plan and country. The day treats those as a look ahead where they are not available on a company account.

Can the training be tailored to our departments?

Yes. Exercises use your own mail, documents and data, so each department practises on its own work. Contact Surendra Pandar to plan a session.

Related reading