USEFUL SOFTWARE / REAL PROBLEMS
I know the problems. AI helps me build the tools.
I don't have a programming background. I start with work I understand, then use AI and a careful process to turn that knowledge into practical apps.
Investing
Timekeeping
Production
Built from firsthand needs
THE PROJECTS
Tools with a job to do.
These aren't demos. They work with real exports, real timekeeping routines, and real production records—while keeping the important decisions with the person using them.
✦
USED WITH LIVE DATA
PORTFOLIO PLANNING
Shows where a portfolio stands, where the plan says it should be, and what could close the gap.
- Imports a brokerage activity export and rebuilds the portfolio.
- Shows allocation, cash, drift, and history.
- Models contributions and rebalancing without placing trades.
HOW IT'S USED
As a private planning console. Any investment decision and transaction still happens outside the app.
Explore the investment app →
✦
RUNNING ON A SPARE TABLET
TIMEKEEPING
Gives a small workplace a straightforward clock-in flow and a clearer path from punches to payroll review.
- Saves PIN-based clock-ins and clock-outs on the device first.
- Keeps working offline and syncs when a connection returns.
- Supports roster, status, payroll reports, and CSV handoff.
HOW IT'S USED
Installed on a spare tablet, with a separate administrator view for review and reporting.
Explore the timekeeping app →
✦
IN USE WITH REAL RECORDS
PRODUCTION
Brings formulations, product costs, component records, compliance labelling, and approval into one local workspace.
- Turns formulations and component data into product-ready information.
- Breaks down materials, labour, packaging, overhead, and profit.
- Prepares component, compliance labelling, and approval views.
HOW IT'S USED
As a local app connected to working production records—not a public cloud dashboard.
Explore the production app →
THE APPROACH
The work starts before the software.
AI can help make an app. It still needs a person who understands the problem, recognizes a useful result, and knows which mistakes matter.
01
Know the work
Start with a routine, decision, or frustration you understand well enough to explain clearly.
02
Build one useful path
Focus on the smallest complete workflow that would genuinely make the job easier.
03
Use it and learn
Try it with real routines and information, keep what helps, and be honest about what's still unfinished.
THE DEVELOPMENT PROCESS
A careful way to build with AI.
AI moves quickly. The process keeps that speed useful: every meaningful change has a clear purpose, sensible boundaries, a review, and evidence that it works.
-
01
Define it
Start with the real problem, decide what success looks like, and name what should not change.
-
02
Build it
Make the smallest complete improvement and protect everything outside its scope.
-
03
Challenge it
Review from a skeptical angle. For riskier work, use a fresh reviewer to look for what the first pass may miss.
-
04
Prove it
Check the finished result—and, when it matters, try it on the real device or in the real routine.
↘
Fewer surprises
Clear boundaries and a second look catch problems before they become part of the routine.
◎
Honest progress
“Built,” “checked,” and “used for real” stay separate, so unfinished work remains visible.
+
Safer momentum
Decisions and evidence carry forward, which means each improvement starts from what is already known.
WHAT “WORKING” MEANS
Clear claims, no theatre.
I separate what an app can do from what has actually been used. Private names, account details, payroll records, and business data stay private.
IN USEThe app has been tried with the real device, workflow, or information it was made for.
OBSERVEDI saw the result directly—on screen, on the device, or in the saved output.
STILL OPENAnything not yet rehearsed or relied on stays clearly labelled as unfinished.
FIELD NOTES
What I learn along the way.
The products come first. These shorter stories cover the surprises, trade-offs, and lessons behind them.
PLANNING APP / REVIEW
A planning app looked finished. A fresh review found four practical problems hiding in less obvious situations.
Read the field note →