Prompt
You ask, it answers — the chatbot era everyone knows.
Hi, I'm Louis — founder of TechSpace
I build AI into the systems real businesses run on.
How I've come to see AI after years of building it into real systems — not the headline version, the working one — and why it matters most for schools.
Scroll
How we got here
How AI got here
It arrived in three distinct stages.
Automation
It takes over the repetitive work, running in the background.
Agents
It reads, decides and acts toward a goal — with a human in charge.
From niche to normal in a year: ~92% of students used AI in 2025, up from two-thirds the year before.
Where we are now
The hardest part of AI isn't the technology — it's knowing where to start. We worked that out on ourselves first and went fully AI-native: every part of the studio now runs on AI and automation. Now we help other organisations do the same — putting AI to work across their people, or standing up an AI workforce alongside them — as their hands-on partner: a part-time CTO and AI change team in one. It's the same three-step pattern in every vertical:
Build the platformadd automationdeploy agents
Seven verticals in — and, honestly, still early. We're working to set a new standard as we go.
The data
AI is already in the classroom
Ready or not, the data is unambiguous:
~92%
of students used AI in 2025 (66% the year before)
83%
of K-12 teachers use AI — weekly users save about 6 hours a week
~2x
Harvard, 2025: students learned about twice as much per hour with an AI tutor
+2σ
Bloom's classic result: 1-to-1 tutoring lifts outcomes two standard deviations — AI finally makes that scalable
1 in 10
schools has any AI policy yet — mostly, no one's steering
Sources: student & teacher AI-use surveys, 2025 · Kestin et al., Scientific Reports (Harvard), 2025 · Bloom, 1984 · UNESCO.
Under the hood
So what's driving all this?
Behind most of what you've just seen sits one idea — the agent: AI that reads, decides and acts, with a person in charge at the end. Here are four I've built, step by step.
A few examples
A few from the real world
The same idea — AI absorbing the mundane heavy lifting — across very different industries.
In numbers
Evidence, not a scoreboard
Rounded figures from the work above — the kind of scale where doing it by hand stops being possible.
480K+
event registrations handled by one media engine
70+
bespoke event microsites built for global brands
5
separate clinic systems unified, fully PII-free
1,252
KL office buildings mapped into searchable data
17 / 3
offence types across three statutes, clamped by law
RM 70M+
in F&B sales processed & analysed for one restaurant group
390K+
sales leads worked through the pipeline
Figures are rounded and unattributed — shown as evidence of scale, not as a leaderboard.
The future of AI
Where does this go?
My read on it, in four points —
It's not a magic box.
Pointing it at a problem doesn't "cure cancer." That isn't what's happening.
It's brilliant at the mundane.
McKinsey estimates today's AI could already handle 40-60% of work hours — almost all of it the repetitive kind.
Which frees people for the non-mundane.
Judgment, creativity, care — the work that genuinely needs a human.
And the one thing it can't do…
connect with other people — becomes the most valuable thing we have.
The more machines handle the busywork, the more time is left for teaching, talking and paying attention to people.
About
TechSpace is a small Malaysian studio that builds AI into the systems real businesses run on.