Case studies

What we have built for Malaysian teams, by department, and the hours each one gave back.

Finance

Claims, invoices and the numbers that close the month.

Google SheetsTrip log · September
StaffTripReasonDays
AimanPenangClient visit3
Mei LingJohor BahruSite audit2
RaviKuantanTraining4
FarahIpohClient visit1

Trigger: month end, trip log closed

Google SheetsClaim rates 2026
GradeReasonPer day
ExecutiveClient visitRM 90
ManagerClient visitRM 120
ManagerTrainingRM 80
ExecutiveSite auditRM 100

Aiman is a manager on a client visit: RM 120

ChatGPTPricing each trip
Matching 4 trips to the rate table
  • Aiman · 3 days × RM 120 = RM 360
  • Mei Ling · 2 days × RM 100 = RM 200
  • Ravi · 4 days × RM 80 = RM 320
  • Farah · 1 day × RM 90 = RM 90
Google SheetsClaims · September
AimanRM 360
Mei LingRM 200
RaviRM 320
FarahRM 90
Total to payRM 970

Ready for finance to approve

33hrs saved a year

From 3 hours a month down to 15 minutes.

Agentic workflow

Outstation claims, priced on their own

The problem
Every staff member who travels is paid at a different rate, depending on who they are and why they went. Finance priced each claim by hand.
The result
The trip log goes in, the right rate is applied to every line, and the totals come back.

HR

Letters, hiring and the paperwork around people.

Google SheetsSalary review · October
NameNew salaryFrom
Aisyah RahmanRM 5,2001 Oct
Daniel TanRM 6,8001 Oct
Priya NairRM 4,6501 Oct
Jason LimRM 7,3001 Oct

Trigger: HR marks the sheet as done

Google DriveTemplates / Salary letter

Salary adjustment

Dear Aisyah Rahman,

We are pleased to tell you that your monthly salary will be RM 5,200 from 1 October 2026.

Details filled from the sheet

ChatGPTChecking the batch
Filling 24 letters
  • Every name matches the staff list
  • Salaries written in full, with RM
  • Dates all set to 1 October 2026
  • Same approved wording in each letter
Google DriveHR / Letters / October 2026
24 files
Salary letter · Aisyah Rahman.pdfJust now
Salary letter · Daniel Tan.pdfJust now
Salary letter · Priya Nair.pdfJust now

24 letters ready to sign

42hrs saved a year

From half a day a month down to 30 minutes.

Agentic workflow

HR letters, issued in one batch

The problem
Name, salary and effective date were retyped into every letter, one person at a time, every month.
The result
HR fills one sheet and the whole batch comes out ready to sign, with the same wording every time.
GmailInbox · careers@
Aisyah RahmanSales Executive9:41 AM
Daniel TanSales Executive9:12 AM
Priya NairSales Executive8:57 AM
Jason LimSales Executive8:30 AM

38 new resumes this week

Google DriveHiring / Sales Executive

What the role needs

  • Must have: 2+ years in B2B sales
  • Must have: own transport, based in Klang Valley
  • Nice to have: speaks Mandarin

Read before every resume

ChatGPTAisyah Rahman · Sales Executive
Checking against the role
  • 4 years selling to factories
  • Lives in Shah Alam, has a car
  • Speaks Mandarin and BM
  • Moved jobs twice in 5 years, fine
Score90/ 100
Google SheetsSales Executive · ranked
1

Aisyah Rahman

Strong B2B, has transport

90
2

Daniel Tan

Good fit, no Mandarin

84
3

Priya Nair

Retail sales only

71
4

Jason Lim

Based in Penang

48

HR picks the shortlist from here

156hrs saved a year

From 4 hours a week down to 1 hour.

AI agent

Resumes, scored and ranked

The problem
HR spent about 4 hours a week opening and reading every resume before anyone could be shortlisted.
The result
The agent scores each resume against the role and says why. HR reviews the ranking and decides the shortlist.

Operations

Forms, reports and the everyday paperwork that keeps things moving.

WhatsAppSite team

Aircon service done, Level 12

4:32 PM

Trigger: photo sent to the group

ChatGPTReading the photo
Pulling out the details
  • Site: Menara Seri, Level 12
  • Job: aircon service, 6 units
  • Done by: Hafiz, 25 Sep 2026
  • Status: all units working
Google DriveTemplates / Service report
SiteMenara Seri, Level 12
JobAircon service, 6 units
TechnicianHafiz
Date25 Sep 2026
StatusAll units working

Template filled on its own

WhatsAppSite team
Aircon service done, Level 12

Service report · Menara Seri.pdf

1 page · PDF

PDF back in the group in a minute

Photos in, finished PDFs out

Agentic workflow
The problem
Details from photos had to be typed into a template by hand before anything could be sent.
The result
Snap a photo and the details fill the template on their own. It comes out as a PDF, ready to use.