Projects and journey
Things built, tested, and still being figured out.
Projects are the practical side of NJL Design Lab: useful tools, AI workflows, financial systems, and household experiments documented as they actually develop.
Shipped and in use
The finished tools live on Tools and Apps.
FI Architect, Deal Analyzer, 1% Better, My Farmers Market Planner, and the Garmin watch face are all live and free. What follows is the workshop behind them: how the work gets done, what is on the bench, and the build log.
The working method
Start with a consequential question.
Make the smallest useful version. Use AI as a collaborator, not an authority. Test the result in the real workflow. Keep the evidence, including the failure, and update the next decision.
01 · Define
What problem is worth solving?
Every project begins with a real friction point, decision, or uncertainty rather than a technology looking for a use.
02 · Build
What is the smallest useful version?
AI helps move from idea to artifact quickly, while the constraints and acceptance criteria stay visible.
03 · Learn
What happened outside the draft?
Projects are updated from actual tests, not from the confidence of the original plan.
Research and experiments
The workshop bench
Some work is a published body of research. Some is infrastructure. Some is paused until the evidence or the tools justify another round.
BuildingAI and household resilience
Research, scenario frameworks, and planning tools about work, money, housing, children, community, and policy in an AI-shaped economy.
Current issue: forecasts and policy examples are time-sensitive, so the work separates evidence from speculation and avoids treating future government support as a household baseline.
Personal AI workflows
Memory, agents, local retrieval, scheduled research, and the safeguards needed to make automation useful without making it opaque.
Current issue: derived indexes can be incomplete or stale; canonical notes and owner-approved memory changes remain authoritative.
AI Visual Inventory Monitor
A camera-independent inventory app that turns images of a pantry, workshop, medicine cabinet, or stockroom into a persistent, confidence-aware inventory using local or cloud AI.
Current state: the image-quality gate, change detection, auto-zoom, persistent inventory, review queue, REST/MCP access, configurable providers, and browser dashboard are implemented.
Open: real iPhone and VoiceOver acceptance, plus the frontier-model accuracy baseline, remain before the next release decision.
Wearable personal interface
An Apple Watch and iPhone system organized around intent instead of app hunting. The first goal is a better interaction model, not a replacement for watchOS.
Current issue: watchOS, battery, and platform boundaries are real constraints. The prototype is paused.
Video and creative production
An AI-assisted film workflow using narration, reference art, generated footage, editing, and continuity checks to explain the future without pretending the tools are effortless.
Current issue: clips can drift in identity, pose, timing, and scene geometry, so the work remains staged behind explicit review and storage gates.
Physical validation successfulBimmer Companion
An iPhone companion app for vehicle telemetry, drive history, diagnostics, and privacy-aware exports.
Lesson: physical validation succeeded, but a low-cost OBD2 scanner cannot provide every metric the app was designed for. Tire pressure is one example; device capability must be detected and explained.
Options Income Tracker
A working tracker for options-income decisions and outcomes.
Next: a free public download and setup documentation will be added in a future release pass.
Financial spreadsheet systems
Three spreadsheet systems originally sold through Etsy are now available as free downloads from the site.
Record: the files are preserved as versioned originals, with the download path now moving from the shop to the public Tools & Apps page.
Project updates
What changed next?
Each update should state the question, the action taken, the observed result, and the next decision. That is how the journey stays useful to someone who was not in the room.
Exploring
The problem is being framed and the evidence is still limited.
Building
A working version exists and is being tested against real use.
Testing
The next decision depends on measurable behavior, not enthusiasm.
Released
The artifact is available, while iteration and maintenance continue.
Paused
The work remains documented, but current evidence does not justify more effort.