Private reflection app
Bottlethought
Send thoughts, photographs, and messages to your future self. Quiet, local-first, and built around the feeling of letting something drift.
HOW I WORK
Learning to work with AI.
Claude Code CLI and Codex have become my partners for ideation, exploration, and development. I still bring the product judgment, taste, and stubbornness; they help me move faster from idea to working software.
That has pushed me beyond web-based SaaS apps into iPhone and iPad experiences. I am learning how to design and develop software for people, and how to work with AI to implement new ideas without losing the human shape of the work.
PUBLISHED APPS
Things you can actually download.
A small shelf of finished iPhone and iPad apps: quiet reflection, tempo practice, beat making, cozy arcade play, and safe cracking by feel. Each one has its own home, support page, privacy policy, and App Store listing.
Tempo-sequence metronome
Metrognomicon
Build hands-free metronome sequences for music with tempo changes, accelerandos, ritardandos, and practice rounds.
Pocket rhythm computer
Drum Daddy
A tactile iPhone drum machine with 16-step sequencing, tap tempo, named memories, track mute and solo, and WAV loop export.
Skill-based prize arcade
Starlight Crane
Aim the claw, collect colorful prizes, and build local Prize Shelves in a warm arcade game for iPhone and iPad.
Tactile safe-cracking game
Three Tells
Turn a safe dial slowly, read the changing rhythm under your finger, find three numbers, and open the safe by feel.
OTHER WORK
More projects.
Prototypes, archives, and AI-shaped tools from figuring out where this new way of building actually belongs.
In Beta
01
Pimola
A calm, private way to split a restaurant bill — scan the receipt, tap who had what, get exact totals. Entirely on-device.
Why I made it
A close friend suggested the idea. It’s a busy market, so I wanted to take a different approach: simple, intuitive interface, and reliable receipt scanning.
What I’m learning
A simple app can feel great if you find the right interaction model and lean into the core purpose.
Bill Splitting · Receipt Scan · On-Device
Visit ↗
In Progress
02
Wonderdeck
An enchanted deck of cards for tiny games, story sparks, and silly prompts — the kind of thing you pull out in a waiting room or at the dinner table.
Why I made it
I wanted the tactile feel of a real, magical deck instead of another list of prompts — cards you flip, fan, and keep.
What I’m learning
The handling — flip, swipe, shuffle — is most of the magic. Illustration and motion carry the rest.
iPhone · SwiftUI · Illustrated
In Progress
03
Clearing
A calm decluttering app that hands you one small zone and one decision at a time, gently biased toward letting things go.
Why I made it
Decluttering stalls on decision paralysis. I wanted an app that works like the executive function you wish you had — narrow the choice, auto-advance, keep moving.
What I’m learning
For an anxious task, the right pace and a single next action beat any feature list.
iPhone · On-Device · Focus
Live
04
PageAudit
AI-assisted website analysis for content, technical, and competitive review.
Why I made it
Most website audits are either too shallow to be useful or too expensive to run often. I wanted to see whether AI could make deeper review self-service, repeatable, and genuinely useful.
What I’m learning
How to structure AI analysis so it produces useful judgment instead of generic recommendations.
Website Analysis · AI Review · Content Strategy
Visit ↗
In Testing
05
Book Club
A reading companion that goes chapter by chapter — talk about what you just read and collect AI-written summary cards as you go.
Why I made it
I wanted to see whether AI could deepen reading without spoiling it: present only for the chapter you’re on, never racing ahead.
What I’m learning
Scoping the AI to exactly where the reader is turns a generic chatbot into a genuine companion.
AI Companion · Reading · Web
Functional Archive
06
Wall St.
A multi-agent stock market simulation with real market data and inter-agent communication.
Why I made it
I wanted to understand how agents behave when they have different roles, goals, access to shared information, and the ability to communicate with each other.
What I’m learning
Agents are most interesting when they have constraints, context, and reasons to disagree.
Multi-Agent · Simulation · Market Data
Visit ↗
Functional Archive
07
F1 Replay
A browser-based replay of past Formula 1 races, reconstructed from public timing data.
Why I made it
I wanted to explore data modeling. I knew the F1 data feeds were extensive. Could they rebuild past races? It turns out: Yes.
What I’m learning
Large data models can be used to produce complex simulations, built in one night.
Simulation · Data Model · Building With AI
Visit ↗
ALSO BUILT
AdjacentPath ↗
An AI-driven questionnaire and report for professionals considering a career change — reframing experience, preferences, and constraints into possible paths.
Drafterly
An AI-assisted content workflow for over-stretched marketing teams. Helps move real work through planning, drafting, review, revision, and publishing while preserving context and voice.
Auto-QA
Automated, highly configurable event and agent driven toolset for testing online web apps.
T-shirts
Automated t-shirt idea generator with detailed text and layout editing tools. Exports to drop-ship vendor formats.
PageAudit Clarity
A streamlined version of PageAudit for visual and content evaluation of a single URL.
Competition Tracker
Monitors and analyzes online communications sources for a user-defined list of companies. Output is a weekly email digest of changes and analysis.
Done By EoD
A lightweight project management tool with an AI project manager as a first-class interface, not a chatbot bolted onto the side.
Site Sight
A tool that gives AI “eyes” on a website as a user experiences it, including visual design, layout, motion, and interaction.
ABOUT
The short version.
I’ve spent 30 years building websites and digital systems for organizations with complicated needs: content-heavy sites, layered stakeholders, governance problems, technical debt, and workflows that rarely fit neatly into a project plan.
That experience gave me a deep respect for the messy reality behind digital work. It also made me hungry for a new chapter.
AI has pulled me back into hands-on building. I’m using it to prototype tools, rethink workflows, and see how far a small, sharp app can go.
This site is where I’m collecting that work: the questions, experiments, and projects from figuring out what comes next.
Say hello.
If any of this overlaps with something you’re thinking about, building, or trying to untangle, I’d be glad to hear from you.