Operational Excellence
People, Process, Technology for Smarter Growth
Fajar Dhirendra Gregory
11 November 2026
Part 1 of 4
Context
What we improve, and where the voices come from.
Continuous improvement keeps raising the new normal from inside
Operational excellence: each and every employee can see the flow of value to the customer, and fix that flow when it breaks down.
Kevin Duggan, IndustryWeek, 2007
Continuous improvement
Small, steady improvements to how work is done, made by the people who do the work.
It comes from inside: speed, quality and waste are ours to improve, whatever the market does.
CI keeps raising it.
Each method fits a different problem
Friction and habits
Frontline workers
Site crew keeps a repair kit at each block
Speed and flow
The whole team
Finished houses stop waiting for handover
Waste and defects
Trained specialists (Belts)
Keys on the promised date, no defects
Defects and variation
Data analysts and statisticians
Why defects repeat, by contractor
Examples are illustrative, set in a housing handover.
Each method created a new normal, and AI is the next one
1951
Kaizen
Toyota, Japan
Habit and culture
Every employee suggests improvements, every day.
40 million+suggestions since 1951, by 2011
Company history
1998
Six Sigma
General Electric
A shared measure
Trained Black Belts run projects on shared measures.
USD 750 million+savings beyond the investment
Self-reported
1999
Lean Six Sigma
Lockheed Martin, LM21
A company routine
Lean and Six Sigma run as one programme across the company.
USD 5 billion+net savings, 1999 to 2004
Company statement
2012
Lean
Virginia Mason, Seattle
How work flows
Any staff member can call a safety alert.
35% → 90%+nurse time with patients
Self-reported
2023 to 2024
AI
Kaiser Permanente, US
People, process, technology
AI drafts the visit note, and the doctor edits it.
15,791 hoursof note time saved in 63 weeks
Self-reported
Years are when each reported result was achieved. Figures are each organisation’s own reports.
Kaizen culture first, then step by step
Start from the bottom
- Don’t start with statistics. It feels like a math test.
- Begin with Kaizen and basic Lean: Gemba walks, 5S.
- Train specialists once spotting waste is a habit.
The first capability to build is listening to the voices
People
- Habit
- Culture
- Skills
- Organisation
- Mandate
Process
- Listening
- Feedback
- Learning
- PDCA
Technology
- Enabler, Support
- Capture, Route
- Show, Analyse
People always come first, for example Kaizen.
Listening, feedback and learning are the sources of Opportunities for Improvement (OFI).
So the first capability to have and to improve is to listen to the voices.
Technology is the enabler and the support: it captures, routes, shows and analyses the voices.
Many unheard voices matter for the future of the organisation
Really meaningful: impactful feedback.
Loud, but we need to say “no” or ignore it.
So soft and subtle that it is often ignored, yet it really matters.
So soft and subtle, but it keeps discouraging us.
We need the capability, in people, process and technology, to listen, capture, understand and process the voices, so signal and noise are clear.
Our own picture of the voices.
The voices show what to improve first
What to improve first
Customer
What buyers, residents and tenants need.
“I need my keys on the promised date.”
Business
What the company needs: profit, growth, compliance.
“We need a 4% higher margin on this estate.”
Process
What our work really delivers today, in data.
“Approving a contractor invoice takes 5 days.”
Market
What competitors, trends and rules are doing.
“A competitor now offers online handover booking.”
Employee
What our people see, suggest and worry about.
“I have an idea to rearrange the key cabinet.”
Kinds of voice say how it sounds. The five voices say who is speaking. Quotes are illustrative.
Every voice needs an owner
Listen only to the business, and morale and quality drop. Listen only to the customer, and the business can go bankrupt.
Every owner needs a path to action
| Voice | Owner | How it becomes action | Example |
|---|---|---|---|
| VOC | Lean and Six Sigma teams | Turned into clear, measurable targets | “Handover is slow” becomes “keys on the promised date” |
| VOB | Executives | Sets money and compliance limits | Show the ROI before buying new inspection tools |
| VOP | Daily operations | Small spike: fix on site. Lasting problem: a project | Defects jump at one block: the site team fixes it |
| VOM | Leadership | Picks which big projects to fund | Buyers move online: fund the digital handover |
| VOE | Site teams | Posted on the team board, fixed fast | A slow checklist, fixed the same week |
Examples are illustrative, set in a housing handover.
We already listen, so now use all five voices
Surveys and feedback, in real time
NPS, CTQ trees
Financial and business reviews
Hoshin Kanri, P&L review
Measure the work directly
Control charts, value stream maps
Research competitors and trends
Benchmarking, trend analysis
Talk with people where the work happens
Gemba walks, Kaizen boards
Small fixes happen on site and big ones become projects
Holding
Executive steering
Picks the year’s top 3 focus areas
Leadership funds major changes
Improvement office
Turns the customer voice into targets
Runs DMAIC projects
Teams fix complex problems
Site huddle
Daily 10 minutes
Small-budget fixes
Quick, low-cost fixes
Sphere size: the size of the issue. Illustration.
Part 2 of 4
Problem
Too many voices to sort, and most never reach anyone.
Too many voices to sort by hand
15 input sources
1,470
voice streams across 98 sites.
98 sites: 70 residential projects, 5 malls, 4 hospitals, 9 hotels, 3 offices, 6 water parks, 1 golf course.
× a voice that reached no one. 1,470 = 15 sources × 98 sites. Lines and × marks are an illustration, not measured traffic.
Most voices never reach anyone
2 in 3
customer problems go unvoiced.
Best and Andreasen, Law and Society Review, 1977. US survey, fieldwork 1975.
27%
of employees in Indonesia are engaged at work.
Gallup, State of the Global Workplace 2026, Indonesia country data (2025 data).
Part 3 of 4
Solution
How AI helps us hear, sort and act on every voice.
AI sorts every voice and people decide
Small and local
Quick Kaizen fix
AI drafts it; the team decides and owns it.
Across departments
Lean project
AI spots bottlenecks before customers feel them.
High variation
Six Sigma project
AI suggests possible causes; the team checks them on site.
For leaders: AI compares market signals with customer complaints, and tracks the results and cost of improvement work.
AI makes listening continuous, but how do we get there
Surveys a few times a year
Real-time analysis of complaints and reviews
Annual planning
Ongoing forecasts of improvement results and cost
Manual time studies
Live process data with early warnings
Manual benchmarking
Constant tracking of competitors and regulation
Suggestion boxes and yearly surveys
AI groups huddle notes and flags safety risks
The company already uses AI and data for market insight and customer segmentation, with more planned for 2026.
?
AI now builds apps and other AI, and it comes in many sizes
Language models and the rest, large to small
Large language models
Data-centre servers
DeepSeek-V3
Small language models
A phone or laptop
Gemma 3n
Not a language model
Speech, voice, forecasts
Whisper, Kokoro, Chronos-2
75%
of Google’s new code is AI-generated
AI builds apps. Company, 2026
23%
faster: a Gemini step found by AlphaEvolve
AI builds AI. Company, 2025
An open model through an API writes “Explain this spot” and “Ask the map”.
Formulas calculate the scores and the travel times.
Open Buildings, WorldCover, WorldPop, Hansen.
Next: the CI Pipeline App, built in 2 effective days.
Voice tools take days to build, and large IT projects take months
Our voice app and 3D model, built with AI
2
effective days
Working days of my time
10.6hours of AI build sessions
28change requests from Fajar
Large IT projects
16.9 months
average ERP project (Panorama survey)
45%over budget, on average
56%less value than predicted
Large IT projects, for comparison, on the right.
The voices stay at the centre. Small AI tools help people hear them and act.
McKinsey and University of Oxford, 5,400+ IT projects, 2012 (projects over USD 15 million). Panorama Consulting, 2017 ERP Report. Build log, as of 6 November 2026.
From scattered voices to one view
- Hear every voiceAll channels, one count
- See where to actSites ranked by what needs attention
- Ask in plain wordsAI answers from the data
Follow one voice from message to lesson
Open until 16 November 2026. All data in the app is invented.
AI helps us listen, innovate and keep raising the new normal
AI helps us listen
15 min
for 6,000 patient comments, against four days by hand. Now used in nine English NHS Trusts.
Imperial BRC, 2025, self-reported
AI helps us innovate
27%
of AI-assisted work would not have been done without AI, in one company’s survey of 132 staff.
Anthropic, 2025, self-reported
AI keeps raising the new normal
and round again
AI empowers, supports and assists us to listen, to innovate and to keep pushing the new normal.
Many small AI tools and a few builders
Build one massive AI system?
Many small AI tools on one spine.
Already running in Indonesia: GoTo says AI agents handle about 35% of complaint and enquiry tickets (company statement); at CIMB Niaga, an AI agent supports service staff in real time.
Make 5,000 people AI-savvy?
A few builders. Everyone a user.
Builders
- Software engineering
- Product and UI/UX
- Design specs
- Team setup
- Self-learning
Everyone: uses the tools and keeps giving their voice.
Tool names are examples. Indonesia examples: company statements listed on AURION (aurion.semai.id).
A small core supports a line that owns the results
coaches
builders join when needed
Communities share what works: belts, builders, voice owners.
The company already runs risk this way: a Risk Management Unit in every business unit, linked on one central platform.
Model based on published practice.
Choose a model for each job, as you choose a ride
One choice, kept for years.
Our voice tools stay plugged in. The model behind the socket can change any time.
Pick a ride for each trip
Sort thousands of voices
Claude Haiku, Gemini FlashDraft a reply to a complaint
Claude Sonnet, GPTFind a root cause across sites
Claude Opus, Claude Fable, Gemini ProRun on our own servers
DeepSeek, Qwen, Llama, MistralGeneric picture of a ride-hailing app. Model names are examples.
Faster apps make improvement compound
Before AI
- Large IT projects take months.
- Few listening channels go digital.
- Many voices are never heard.
- Few improvements.
The compounding loop
- 1Faster apps
- 2More digital listening
- 3More voices heard and acted on
- 4More improvements
and round again
After AI
- A small app took us 2 effective days.
- Every listening channel can go digital.
- Voices are heard, sorted and acted on.
- Many small improvements.
Start with one small improvement this month.
Part 4 of 4
Close
What to take home.
Listen to every voice and keep raising the new normal
- 1Operational Excellence and Continuous ImprovementA baseline we keep raising, from inside.
- 2People, Process, TechnologyPeople come first.
- 3Listening, Feedback, LearningThe first capability to build is listening.
- 4The AI LandscapeAI builds apps and other AI, in many sizes, and an app now takes days.
- 5AI and the New NormalAI helps us listen better, be more innovative and keep pushing our new normal.
And smarter growth starts from inside.
One voice, one small fix, next week
Week of 16 November 2026
Pick one voice
you heard this week, in your next huddle or team meeting.
With AI: AI helps group the voices you have.
Agree one small fix
with the people who do the work.
With AI: AI suggests options, and people decide.
Share what happened
so others can learn from it.
With AI: AI drafts the short summary.
With AI: only tools your company approves, and no personal data.
This week: try the demo app. Next week: one voice, one small fix, shared.
For an algorithm, an error is a flaw to be corrected; for a person, however, an error can be a catalyst for profound change.
Pope Leo XIV’s first encyclical, and it is about AI. An encyclical is one of the Pope’s highest-ranking teaching letters. Signed on 15 May 2026. It starts from the human person. Its subtitle: “On safeguarding the human person in the time of artificial intelligence.” Presented in the Vatican on 25 May 2026, with Chris Olah, co-founder of Anthropic, among the speakers.
Thank you
Fajar Dhirendra Gregory