Practical tools for the parts of work that keep getting stuck.
I’m Os Ishmael. I’m building 2nd Bell, AI Skills Tutor HK, Qithym, Loamnote, ActaOS, and an early robotics data venture.
In short
Start with the handoff, habit, or decision that breaks. Then shape the simplest useful product around it. AI helps when it makes the work clearer.
Current portfolio: 2nd Bell, AI Skills Tutor HK, Qithym, Loamnote, ActaOS, and Robotics Data Venture
Current systems
Four tools and two ventures built around real bottlenecks.
Each one starts with inputs, outputs, and a clear human job.
Revenue workflow system
2nd Bell
Input: paid enquiries. Output: fewer lost leads. The tool tracks replies, follow-up, bookings, quotes, and missed handoffs.
Built with Next.js
Visit 2nd Bell →Learning system
AI Skills Tutor HK
Input: a student task. Output: better research, checking, and project habits. Students learn how systems think.
Static HTML, video, structured schema
Explore AI Skills Tutor HK →Productivity
Qithym
Input: a work session that lost shape. Output: a small next step. It helps people restart after breaks.
Built with React and Vercel
Follow the build →Focus sound tool
Loamnote
Input: a noisy room or scattered focus. Output: local soundscapes, timers, and Pomodoro cycles that stay out of the way.
Built with React, Vite, and Web Audio
Open Loamnote →The common thread
AI is a utility, not the pitch.
Each project starts with a real operating problem.
Businesses lose leads after first contact.
Students use AI without knowing how to check it.
Workers lose rhythm after breaks and context switching.
I build small systems around those gaps.
How I build
Clean systems before clever tools.
Find the constraint
Where does the work slow down, leak, or become unclear?
Map the workflow
What happens next? Who owns it? What should the software make obvious?
Use AI as a tool
Use it for narrow jobs: draft, classify, summarize, or check.
Featured venture
2nd Bell shows where paid leads get lost.
It follows the path from enquiry to reply, booking, quote, sale, payment, and repeat work.
The goal is simple: show the leak and make the next fix obvious.
AI Skills Tutor HK
Practical AI lessons for students and families.
AI Skills Tutor HK teaches students how to use AI without outsourcing thinking.
The focus is checking, explaining, and knowing when human judgement matters.
Writing and build notes
Notes from the work in progress.
I write about what I build, test, ship, and learn.
Read the notesWork with me
Choose the path that fits the problem.
For service businesses
Find where revenue is getting stuck after demand is created.
Explore 2nd BellFor website projects
Build a clearer site for a Hong Kong SME that needs trust and enquiries.
Explore Website HKFor schools and parents
Teach students to use AI, check it, and explain their thinking.
Explore AI Skills Tutor HKFor robotics data ventures
Discuss the thesis, validation work, and structured real-world task data for robotics and embodied AI teams.
Explore the ventureFor investors and partners
Talk about products, markets, and small software bets.
Start a conversationFor advisory work
Bring operator judgement to workflows, software, and commercial clarity.
Work with me