Shipped four approval-console tickets to the always-on box in one day
Every ticket survived three adversarial critic rounds with real blocks before it went live.
4 live ships, 12 critic rounds
I spent 10 years running small businesses: a gym, coaching practice, content operation. Then I discovered I could build with AI. Now I run an AI automation consultancy and a paid community of builders.
Started health and fitness coaching
Opened Lift Life Fitness gym
Published "Health Without Sacrifice" book
Discovered Claude Code, shipped first app
Founded Zero to Automated
Became a verified mentor in the Vibe Marketing Community (2,500+ members)
Launched the Zero to Automated community (100 members in the first 3 weeks)
Built 10+ production CLI tools and a full content pipeline
Flipped ZTA to paid: first paying member on day one
"I went from fitness coach to running an AI automation business with paying clients and a community of 200+. Not because I learned to code. Because I learned to build with AI."
Want to work together?
Let's TalkA running record of what I ship: builds, client work, writing, and milestones. Newest first.
Every ticket survived three adversarial critic rounds with real blocks before it went live.
4 live ships, 12 critic rounds
A deal that went cold for nine weeks came back. I walked into their office and left with a signed assessment.
Signed on site, scoped the same day
Two mockups got rejected. The third was built desktop-first from panels and computed styles measured off a live reference app, read-only.
Three tickets shipped to the box the same week
Ten tickets, every one run past a critic before it could ship, and roughly ninety root-cause fixes landed before night one.
~90 root-cause fixes before night one
Pitched an AI advisory system built for a financial-advisory firm live on stage as a finalist in the Amazon Web Services AI Pitch (Indiana SBDC, Valparaiso).
From a competitor teardown to a fully autonomous daily pipeline in two days: news sweep, de-slop-gated script, cards, cloned voice and music, render, avatar bubble, upload and schedule with read-back.
11 Shorts in four days, the last two with nobody at the keyboard
A credential broker that types secrets from 1Password without the value ever entering the container, a closed nine-operation browser vocabulary behind an origin allowlist, and a readiness ladder proven green twice.
Readiness ladder R1 to R7, green twice
The AI-slop scanner, 26 rules with the em-dash rules folded in, went from a members-only Vault skill to a public repo anyone can install.
26 rules
A two-arm A/B, identical agents on identical work with only one arm's guardrails removed, against a sealed failure criterion and seven sealed predictions. The criterion fired and all seven predictions were wrong.
0 of 7 predictions held
The agent server now serves its own decision console over Tailscale with TLS. Findings, drafts and approvals flow through one recorded writer, and the first real approval was executed end to end from the console.
10 of its own bugs caught by its own checks before launch
Rebuilt browser automation around trusted input after measuring the blind spot: synthetic clicks land but are stamped untrusted, and a defended page drops them with no error to debug.
Trust gate proven 6/6
A Vault skill that revives cold leads and writes the follow-up ladder, and structurally cannot send: the packer fails the build if a send path appears anywhere in the skill.
30/30 tests green from inside the packed zip
After a rotation silently invalidated a recovery bundle, the local env file is now rendered from a 1Password vault instead of hand-edited. 109 variables seeded into 44 items, verified 109 of 109 match.
109/109 match, first pull changed nothing
The judgment layer above the em-dash remover, split into three tiers by determinism rather than topic, then tested head to head against a blank model on the same draft.
12/12 planted tells caught, 0% of a clean draft changed
A prospect in the insurance space opened with the standard objection, that AI takes away the personal relationship. By the end of the call there was a working dashboard demo on screen.
22 fictional clients across 5 plan types, built during the call
For a financial-advisory firm's client-facing AI build, the model vendor declined a custom zero-data-retention agreement. The answer was AWS Bedrock with retention set to none: zero retention by default, zero operator access, self-serve, no platform fee.
A member-engagement system for a community platform with no public API: campaign batching, weekly digest bundles, and hard guardrails that abort the whole batch on the first unverified send.
186-member campaign run to exhaustion, self-terminated by design
Members now pull branded skills straight from a private GitHub repo via a collaborator invite instead of downloading a zip. Version-controlled, updatable, revocable per member.
A reimplementation of the Dreams API curation pattern: extract, gate, single-model judge, then a deterministic script that re-verifies every evidence quote verbatim and enforces a two-session promotion gate.
30 regression cases plus eval cases
A 15-beat, Wallace & Gromit-style clay-world sales video, live on the Zero to Automated About page: script → AI voiceover → AI-generated living-clay footage → vector composition, built beat-by-beat under human review.
15/15 quality gates passed
A skill that pulls fresh funding, price, and volatility for any ticker and bakes it into a single offline HTML dashboard. Shipped to the free ZTA Vault.
Funding rates on stocks, FX, commodities, and indices, not just crypto
Two alternating post types, proof posts and news takes, each gated through a 13-persona focus group before it ships, funneling to the free ZTA Vault.
13-persona focus-group gate on every post
A brand-config skill that turns raw business inputs into branded proposals and invoices (HTML to PDF, no external service). Built for the free ZTA Vault.
Runs the Brain on a Windows box, reachable by SSH from anywhere over a key-only, Mac-locked Tailnet.
An interactive prototype for a dental practice weighing an AI patient-intake system: a working demo with an auto-playing narrated walkthrough that pitched itself, and moved the engagement to scope and pricing.
A live inbound AI receptionist ("Summer") for an insurance agency: state-law AI disclosure, structured medication read-back, and a warm hand-off to a human.
Live AI receptionist with human hand-off
An advisor dashboard with auto-generated quarterly PDF explainers and a compliance gate requiring a verbatim source quote and provenance tag on every output.
Launched VibeVoice, my native macOS voice-to-text app: code-signed with a hardened runtime, automatic updates via Sparkle, shipped as a paid product.
A deep audit of a 137,000-line production platform: built an AI-to-AI discovery protocol, surfaced silent backup failures and a runaway AI-spend leak, and designed a zero-downtime fix with a phased backlog.
105-item phased remediation backlog
Verified mentor in the 2,500+ member Vibe Marketing Community, and presented the AAA Authority Acceleration Framework to the full community.
2,500+ member community
Launched the AI automation consulting practice (now Zero to Automated), the origin marker of the current chapter.