Hi, I'm John Davidson.
I've spent 15 years building products in complex, data-intensive industries — energy, agriculture, fintech, enterprise AI. I've led teams, scaled platforms, and owned product lines driving $100M+ in revenue at companies including Opower, Capital One, Granular, and Mach49.
For the past two years, I've been building with AI as a primary collaborator — not just using AI tools, but leading AI teammates through real product development. Scoping, reviewing, iterating, shipping. The projects below are the result. The product thinking behind them isn't small.
How I Work With AI Teammates
Most people use AI like a search engine with better grammar. I work with AI the way I've always worked with teams — clear problem framing, strong direction, critical review, and iteration until it's right.
The problem comes first. Always.
Rekindl didn't start with "what can AI do?" It started with my great-grandparents' biography on a shelf and the question of why almost no family has anything like it. Sixes started with me losing billable hours to spreadsheet friction. AI is how I built the solutions. It's never why I built them.
AI output is a first draft from a talented teammate.
It's often 85% right and 15% wrong in the ways that matter. When I'm reviewing AI-generated code or copy, I'm doing the same thing I do managing teams — pattern-matching against what great looks like, catching what's off, and directing the next iteration. The process has changed. The discipline didn't.
I build systems, not prompts.
Rekindl isn't a chatbot. It's conversation flow logic, context management across sessions, QA frameworks for monitoring regressions, and notification routing that surfaces the right story to the right family member. Sixes is a native app with CI/CD and App Store deployment. I've architected real engineering systems through AI collaboration.
I know when to stop building.
Liminary was a coaching app for laid-off professionals. The product worked. Users engaged. Nobody paid. We killed it. Having AI teammates that can build anything in hours makes this discipline harder, not easier — the temptation to keep shipping features instead of facing a market signal is real. Knowing when the answer is "no" is the product skill I value most.
Projects: Proof in Practice
Rekindl — Preserving Life Stories Through AI
rekindl.coMy great-grandparents' biography sits on a shelf in my house — interviews conducted in Russian by my grandmother, edited into English by my aunt. It's one of the most treasured things I own, and almost nobody has anything like it because creating something like it has always required enormous effort.
Rekindl changes that. An AI interviewer conducts warm, adaptive conversations with elderly family members via voice or text, then crafts their stories in their own voice. Family members get notified when new stories are ready, tuned to what matters most to them.
What was hard: The product has been rebuilt from the ground up multiple times as AI capabilities evolved. Each rebuild was not a failure. Each was a response to new technical possibilities and user feedback. I built QA systems for monitoring regressions across sweeping changes to conversational behavior, and designed every interaction for users who aren't tech-savvy — meaning it has to feel effortless and personal.
What it reinforced: Long-term iteration discipline. The hardest thing about AI-first products isn't the initial build — it's maintaining quality and coherence as you evolve the system over months. The same challenge I've encountered time and time again: how do you evolve without breaking what matters most?
Currently in private beta with families.
Sixes — Billable Time Tracking in Taps
Download on the App StoreEvery consultant I talked to had the same problem: losing billable time because tracking small tasks in a spreadsheet created enough friction to skip it. Sixes makes it dead simple — one tap = 6 minutes logged. The same 0.1-hour increment professional services firms have used for decades. No timers, no categories, no accounts. Privacy-first with all data stored locally.
What was hard: The MVP was straightforward though it was my first native app. What surprised me was what happened after launch — users immediately revealed a deeper problem space than I'd designed for. The tool I built to track time became a window into how consultants think about their work and their value.
What it reinforced: Scratch-your-own-itch discovery is powerful, but your itch is just the starting hypothesis. The real product emerges from what users do that you didn't expect.
Randimal — Speed to Market With AI
randimal-ai.comA creative experiment built in 3.5 weeks: combine any three creatures to generate a unique hybrid with its own personality, backstory, and artwork. Payments integrated, 15% to conservation. A deliberate test of how fast you can ship with AI collaborators handling full-stack development, design, and go-to-market.
What it reinforced: Speed is a product skill — not rushing, but knowing how to scope tightly, decide quickly, and ship before you've perfected it. The teams that ship fastest aren't the ones with the best tools. They're the ones with the clearest thinking.
Liminary — Knowing When to Stop
A coaching app for recently laid-off professionals. I built the product, tested the concept, and discovered the audience was willing to engage but not willing to pay. The unit economics didn't work. So we killed it.
What it reinforced: Not every good idea is a good product. The discipline to stop when the market signal is clear — even when you like what you've made — is one of the hardest skills in product management. Liminary reminded me that the best product decision is sometimes "no."
"He didn't just tell us what went wrong on one project — he identified the systemic patterns that would have repeated across future initiatives if left unaddressed... He was direct without being confrontational, and focused on building our team's capabilities rather than creating dependency on his involvement."
Writing
I write about creativity, vulnerability, and what happens when intelligence gets abundant and cheap.
A Sovereign Creator on Substack →Let's chat.
I'm always interested in hard product problems — especially at the intersection of AI capability and real organizational impact. Currently open to product leadership roles where 15 years of building experience and genuine AI fluency can make a difference.
When I'm not building software, I'm in my workshop making furniture and metalwork, training for a half-marathon, or somewhere in the mountains on wheels, skis, or by foot.