Why We Still Run Sessions in a Room
Most of our AI training works fine over Zoom. The in-person sessions do one thing online cannot, and it has nothing to do with the slides.
Read the post →Short notes from AICoach.my - what happened at our events, practical AI tips you can use today, and things worth sharing.
Most of our AI training works fine over Zoom. The in-person sessions do one thing online cannot, and it has nothing to do with the slides.
Read the post →
Making app interfaces from plain text, generating leads with Claude, and optimising content so AI answer engines recommend you. Three talks, one Sunday.
Read the post →
James Chong broke down what actually makes a landing page fill seats: promise the result first, put the form early, and repeat one call to action.
Read the post →
A National Training Week session on the Claude ecosystem, building a working web app live, and publishing it. Over 300 registered and more than half turned up on the night.
Read the post →
Session one covered prompting, fact-checking and artifacts. Several participants had already turned it into real reports and dashboards for their own work.
Read the post →
A big room, a wide spread of roles, and a full day of AI training. The third group from the same company to come through, which tells its own story.
Read the post →
A team of marketers, copywriters and social media managers spent a day on prompting, image generation and building their own AI tools. The pace was quick.
Read the post →
Past participants from several programs came back for one evening to see what Claude does differently, and how to bring their AI memory across with them.
Read the post →
Demo Day #18 ran through a currency tracker, a burnout monitor, a property defect form and eight more, all built by people with no programming background.
Read the post →
A company brought us in for two full days on Claude AI. Two groups through the same boardroom, and a lot of work that will not be done by hand again.
Read the post →
What Claude can actually do, how to automate WhatsApp with it, and how to get more done using Claude Skills. Then everyone built something.
Read the post →
A session with an asset management team on letting Claude handle inbound enquiries, log them to a spreadsheet, and start conversations. No code involved.
Read the post →
The pioneer group started with projects, custom skills, and the browser extension. Being first means nobody can tell you how it usually goes.
Read the post →
A hands-on hour installing an AI add-in for Excel, building dashboards, and running a real salary equity analysis on a school administrator's data.
Read the post →
A bilingual business diagnostic, a financial ratio analyser, and a family planning tool - the apps three participants built to complete the Google AI Program.
Read the post →
Three participants from Groups 19 and 21 presented working apps: a health planner, a decision compiler, and a platform matching parents with tutors.
Read the post →
Two talks on why a language app keeps people coming back every day, and what small businesses can borrow from it. Then three hours of building.
Read the post →
Three talks, then everyone opened a laptop and built. A calendar app, a feedback tool, and a website - all made in one Sunday afternoon in January.
Read the post →
The DIY Your App & Website with AI workshop: write a product requirement document, build a functional app with no coding, and share it for people to try.
Read the post →
Three talks from Yee Peng, Li Hong and Ricky Soo, then a room full of people building nutrition assistants, speechwriting tools and an accounting website.
Read the post →
An hour on Google's Gemini: generating and editing images with Nano Banana, and building Gems, the small custom assistants that save repeat work.
Read the post →
Team accounts, prompt engineering, and Custom GPTs, run twice in a day for management and departmental staff at a corporate client.
Read the post →
An AI agent can answer customer questions, draft content and watch your data overnight. Start by naming the one job you want it to own.
Read the post →
The old website chatbot answered five questions and gave up. An AI chatbot reads your own material and answers in your own words instead.
Read the post →
The tasks worth automating are rarely the big ones. They are the small repeats nobody notices until you add up the hours they quietly take.
Read the post →
The first Inno Day packed Keep In Touch Co-Working Space for over three hours of building websites, AI agents and automations from scratch.
Read the post →
AI is poor at knowing what you want to say and good at getting it said. Used in that order, it turns a blank page into a working draft.
Read the post →
A form does not have to just collect a name and sit there. With AI behind it, it can answer, qualify, book and follow up on its own.
Read the post →
A chatbot with a personality is not a gimmick. Character shapes how people talk to it, and how much they are willing to tell it.
Read the post →
Someone floated the idea of an AI that reads palms. Three months later it was live, with a seven-episode video story attached.
Read the post →
The useful assistants are not the ones that think for you. They are the ones that handle the admin around the thinking.
Read the post →
Good ads are usually the tenth version, not the first. AI matters because it makes getting to the tenth version cheap.
Read the post →
People remember how a brand made them feel long after they have forgotten what it claimed. That feeling is something you can work on deliberately.
Read the post →
Brainstorm with AI, then let a design tool turn the content into slides. The point is not speed for its own sake - it is getting a draft to react to.
Read the post →
The hardest part of a first product is not building it. It is finding out whether anyone wants it before you have spent six months.
Read the post →
One line from a participant sums up the change better than anything we could write. Previously the non-tech could not do anything. But now they can do it.
Read the post →
A spreadsheet, a writing tool and a connector are enough to keep a posting schedule running without anyone remembering to post.
Read the post →
We expected the technical sessions to be the memorable ones. Some of the tools people liked most were the ones that simply said something kind.
Read the post →
One prompt trying to do five jobs usually does none of them well. Splitting the work between several agents is what changes the output.
Read the post →
Most business software works fine alone and badly together. Connecting them is usually a bigger win than adding another tool to the pile.
Read the post →
How you say a thing shapes what comes back, whether you are talking to a person or to a model. Vagueness in, vagueness out.
Read the post →
Naming it, writing the instructions, choosing the conversation starters. Building a Custom GPT works better as a group activity than a solo one.
Read the post →
One business owner automated about 80 percent of their e-invoicing work with AI. Less data entry, faster payments, and it is not only for big companies.
Read the post →
A good chatbot is not clever. It is the receptionist who greets everyone, answers the ordinary questions, and passes on the ones that need you.
Read the post →
What made Iron Man's assistant appealing was not the technology. It was that it anticipated, decided and acted without being asked twice.
Read the post →
A blog post, a Facebook post and a matching image, produced end to end without anyone touching a keyboard. The first run was rough. It got better.
Read the post →
Building a joke-telling GPT sounds like a waste of an afternoon. It teaches more about tone, instructions and taste than most serious projects.
Read the post →
The goal we keep coming back to is unglamorous. Give people who have never written a line of code the ability to build the thing they need.
Read the post →
We invented a law firm that does not exist, built its website, and gave it a chatbot. The point was to see how far an afternoon goes.
Read the post →
Flowcharts usually mean an hour of nudging boxes into alignment. Describing the process in plain text and letting the diagram draw itself is quicker.
Read the post →
Most nerves before a talk come from not knowing whether the thing works. AI will not stand on the stage for you, but it will listen first.
Read the post →
Someone wanted the news that matters to them and nothing else. Getting there turned out to be less about fetching stories and more about choosing sources.
Read the post →
Getting AI to write an article is old news. Getting it to review the article against what search engines reward is the more useful trick.
Read the post →
A business report used to be a week of gathering and a weekend of writing. We built a version where you fill in a form and read the result.
Read the post →
Not everyone can produce a good pantun on demand. One participant built something that can, and it turned a chore back into a pleasure.
Read the post →
The same technology that makes it easy to produce convincing nonsense also makes it faster to check. Which side wins depends on how you use it.
Read the post →
A bot that listens to whatever is on your mind, replies with something useful, and emails you a summary afterwards. Built inside an app you already have.
Read the post →
A landing page is not a smaller website. It has a single job, and most of the craft is in removing everything that gets in its way.
Read the post →
The most useful way to picture AI in a small business is not a genius. It is a capable new hire who never gets bored of the boring parts.
Read the post →
The interesting part of AI in education is not the technology. It is what a teacher with twenty years of experience does once the paperwork shrinks.
Read the post →
Building a website used to mean HTML or a budget. In one session, people went from a topic to a published page in about ten minutes.
Read the post →
Some chatbots are a menu with extra steps. The difference between those and the useful ones is almost entirely what you feed them.
Read the post →
A session that turned into a debate. AI can produce endless variations of a thing, but variation is not the same as having something to say.
Read the post →
Checking prices by hand every morning is a habit, not a job. One session replaced it with a sheet that fills itself before anyone arrives.
Read the post →
A demonstration of Google's Gemini and what changes when the AI can see and hear rather than only read what you type.
Read the post →
Generative AI for content creation, run for Group 4 in January 2025 - building the pages a small business actually needs, without a designer.
Read the post →
Handling enquiries, capturing leads and answering in more than one language - the session covered what a chatbot actually changes for a small business.
Read the post →
Prompt engineering ran twice that winter, for Group 4 and again for Group 1. It is the first module for a reason - everything after it depends on it.
Read the post →
An introduction to Malaysia's free national AI literacy course, run as a live session with a quiz. The response was better than expected.
Read the post →
Basic workflows with Make.com, then connecting apps so they talk to each other. No code, and the sessions ran across three groups that winter.
Read the post →
An online session on using AI to lift a presentation - not just to generate the slides, but to sharpen what you actually say.
Read the post →
Two features that sound similar and do different jobs. Live demos of when to reach for a Custom GPT and when a Project is the better fit.
Read the post →
First a Custom GPT from scratch, then connecting it to outside systems. Beginners with no coding background built working tools in a single session.
Read the post →
Custom GPTs for language learning and test prep, then a session where children learned to spot AI-generated content through games.
Read the post →