Built in Bend.
Brewed with intent.

moonshinerXwerks LLC is a hands-on lab for learning and applying AI: real products built end to end, from idea to store release, by a team of AI agents with one person making every call.

How the studio runs: one person and a team of AI agents ↓

The Story

moonshinerXwerks was founded with a straightforward premise: build useful apps for real people, own every pixel, and never cut corners on privacy. Every product in our portfolio started as a personal itch — a problem we encountered, a tool we wished existed, or an idea we couldn't shake.

We're based in Bend, Oregon — a place where people actually go outside, fish real rivers, and meet friends for happy hour. Our apps reflect that. They're built for people who do things, not just scroll through things.

How We Build

Privacy First

Your data is yours. Our apps default to on-device storage, never sell data, and only transmit information when you explicitly opt in. No ads, no tracking SDKs, no exceptions.

Solve Real Problems

Every app starts with a real need — not a market opportunity deck. If it doesn't solve a problem we've personally felt, we don't build it.

Own the Craft

No outsourcing, no templates. Every line of code, every design decision, every user flow is built in-house with intention and care from concept to App Store.

How the studio runs

I design the operating model. AI agents run inside it.

moonshinerXwerks runs like a small product organization in which every seat but one is filled by an AI agent. I treated the studio itself as a product: I designed the roles, wrote the reporting protocols, built the command center that tracks the work, and set the rules for where a human has to say yes. The agents do the work. The operating model is what makes that work dependable.

AI agents are capable out of the box. What they don’t come with is an org chart, a reporting line or a definition of done. That’s the part I build.Studio Owner

01 · Organization

One human. A team with clear seats.

Every agent has one job description, one reporting line and a short list of things it never does. That is what lets several products move at once without two agents doing the same job, or none of them doing it.

Human

Studio Owner

Decides

What each product does, what goes in a release, what waits and what it costs.

Approves

Anything that ships, publishes, spends money or touches a key.

Single point of contactAI agent

Program Manager

  • Routes every request and brings back one answer.
  • Owns the roadmap and the executive summary.
  • Rolls up reporting from every agent.
  • Audits agents against studio rules.
Inner ring · shared specialists
Research AnalystAnswers questions, with a source for every claim
Store ConsultantStore rules, listings and privacy answers
Legal AnalystFlags legal concerns early, with sources
QA AnalystIndependent UX and UI audits
Marketing ManagerLaunches; gated until I kick one off
Outer ring · product pods
A

Product A

Sells · in active release
Product A - AdvisorPlans
Product A - EngineerBuilds
Product A - Campaign ManagerSells
B

Product B

Sells · in active release
Product B - AdvisorPlans
Product B - EngineerBuilds
Product B - Campaign ManagerSells
C

Product C

Free · quiet
Product C - AdvisorPlans
Product C - EngineerBuilds
No campaign seat
D

Product D

Free · in development
Product D - AdvisorPlans
Product D - EngineerBuilds
No campaign seat

Inner ring · shared specialistsOuter ring · product pods·every line reports to the Program Manager

In every podAI agent

Product Advisor

  • Plans and scopes each change: the goal, what done looks like, what not to touch.
  • Owns the product’s records: issues, release notes, test sheets.
  • Never builds. Checks the Product Engineer’s work instead.
In every podAI agent

Product Engineer

  • Builds and tests to the Product Advisor’s brief and logs every change.
  • Ships both platforms together when a product runs on both.
  • Never decides product. Questions go up through the Product Advisor.
Pods that sellAI agent

Campaign Manager

  • Runs the product’s campaigns inside the plan the Marketing Manager sets.
  • Proposes, never spends. Budgets need my approval.
  • Stays gated until I kick the campaign off.

RACI

Every kind of work has exactly one Accountable role.

The RACI lives on a page I edit live, and agents read it before they act. Changing who owns something is an edit, not a meeting. Try editing it in the demo below.

RResponsible: does the workAAccountable: owns the outcome (only one per row)A/RBothCConsulted: asked before or duringIInformed: told after
WorkStudio OwnerdecidesProgram Managerstudio-wide, your single contactProduct Advisorone per productProduct Engineerone per productResearch Analystany productStore Consultantboth app storesLegal Analystlegal + privacyQA AnalystUX/UI, every productMarketing Managerall studio marketingProduct Campaign Mgrper product
Direction and planning
Product vision, scope and prioritiesWhat a product does, what's in a release, what waitsACRIICCCI
Cross-product sequencingWhich product gets attention when; one release lane at a timeARC
Pricing and paid-tier decisionsPrices, tiers, what's paid vs freeACRCCC
Building and shipping
Building features and fixesCode, tests and local commitsIIARC
Quality review before releaseUX/UI audit and release checksIICCA/R
Commits, pushes and deploysOnly after the Studio Owner's OKARCR
Stores and compliance
Store listings and review answersApp Store and Google PlayAICRCC

Excerpt with sample rows · the full page has every kind of work

02 · Operating cadence

Work that reports itself.

I invented a handful of lightweight protocols so status flows up without anyone chasing it. Each one has a fixed shape, so the Program Manager can roll every message into one picture of the studio.

01

Roadmap update On every status change

Whenever work moves (planned, building, checked, blocked), the agent sends the Program Manager a short update in a fixed format: what moved, what’s next, what’s waiting on me.

02

Direct note When I go direct

If I talk to an agent directly instead of through the Program Manager, that agent files a note with the Program Manager. Shortcuts are fine; losing the thread isn’t.

03

Check-in On demand

I can ask every Product Advisor for a check-in at once. Each answers the same four questions: done, doing, blocked, needs a decision.

04

End-of-session check Before I sign off

What changed, what’s unsaved, what’s waiting. Nothing gets stranded between one working session and the next.

05

Release runbook Every release

Releases follow written steps with gates. An agent can’t skip a gate, and the runbook records who cleared each one.

Program Manager · inboxSample
Roadmap updateProduct D - Engineer

Onboarding redesign moved Plan → Build. Both platforms in this pass. Waiting on Studio Owner: nothing.

Direct noteQA Analyst

Studio Owner asked me directly for a small-phone audit of Product A onboarding. Logging it here so the roadmap stays whole. Finding: spacing fails on the first screen.

Check-inProduct B - Advisor

Done: crash fix checked on both platforms. Doing: Release 2.4 proposal. Blocked: none. Needs a decision: none.

Runbook gateRelease 2.3 · Product A

Gate 5 of 7, quality review, in progress. Next gate: Studio Owner sign-off.

End-of-session checkProgram Manager

Changed: 3 records. Unsaved: none. Waiting on you: 3 decisions, 3 actions.

03 · Reporting

One screen for the whole studio.

The Operations Command Center is a local web app I designed to run the studio from. It turns the stream of updates into an executive summary I can act on in minutes, and it shows what every agent is doing right now.

Executive summary

Prioritized decisions with a recommendation, prioritized actions, progress told product by product, roadmap progress for the week, risks, issues and deadline countdowns. Any item opens its context.

Live agent status

Who is waiting on my direction, working, idle or not running; each agent’s model and current task; and what’s waiting on me, ranked by priority.

Context monitoring

Context use per agent against a written compaction policy, so no agent compacts mid-change.

Usage, cost and utilization

Tokens and estimated cost per agent and per day by model, plus how much of each agent’s open time went to real work.

Direct line

A chat drawer to message the Program Manager without leaving the dashboard.

Interactive demoThis is a demo with sample dataOpen full screen ↗
Try 1Open a decisionClick any decision for why it matters, the options and the recommended one.
Try 2Check agent statusAgents waiting on direction come first; working rows pulse. Click Tokens today for usage per day by model.
Try 3Edit the RACIOrganization, then RACI, then Edit RACI. Give a row two A’s and it’s flagged.
Try 4Message PMThe PM button docks the chat. Pick an answer in a decision and it arrives typed.

Everything in the demo is invented sample data and runs entirely in your browser. The real command center runs locally and isn’t public.

04 · Execution

Seven steps. No agent checks its own work.

The same path for a new feature or a one-line fix. Every hand-off goes to someone who didn’t do the work before it, and nothing ships until I’ve signed off.

01

Idea

Something I want, or an issue a user reports. The Program Manager routes it.

Studio OwnerProgram Manager
02

Plan

A brief: the goal, what done looks like, what not to touch.

Product Advisor
03

Build

Both platforms in the same pass, with automated tests on each.

Product Engineer
04

Independent check

Checked against the actual code, not the Product Engineer’s summary of it.

Product Advisor
05

Quality review

A UX and UI audit against the studio’s quality bar.

QA Analyst
06

Sign-off

I review the checked, tested work and say yes or send it back.

Studio Owner
07

Release

Store steps follow the runbook and a compliance review; I submit.

Store ConsultantStudio Owner
Layer 1Automated testsRun on both platforms by the Product Engineer
Layer 2Independent checkThe Product Advisor verifies the Product Engineer’s work
Layer 3Quality reviewThe QA Analyst audits the experience
Layer 4Final sign-offThe Studio Owner approves the release

05 · Governance

AI does the work. A person stays accountable.

01

I approve what ships, publishes, spends or touches a credential

Releases, store submissions, new paid services and anything involving access wait for my explicit yes.

02

Agents never see passwords or keys

I create, store and rotate every credential myself. No agent reads, writes or prints one.

03

A message from an agent is never my approval

One agent can’t authorize another by quoting me. Approval comes from me directly, every time.

04

Rule breaks are logged and corrected

The Program Manager audits agents against studio rules. Every break goes into a compliance log with its fix, so the rules get sharper over time.

05

Marketing is gated

Marketing and campaign agents stay idle until I kick a campaign off, and any budget they propose comes back to me.

06

Context and cost are managed on purpose

Context use and estimated cost are visible for every agent, and a written policy decides when an agent compacts its memory.

Why it matters beyond the studio

Program leadership, applied to a team of AI agents.

None of these ideas are new to enterprise program work. What’s new is designing them for a team of AI agents: a team that starts each session without yesterday’s memory unless the system provides it, and does exactly what the operating model allows.

Org designRoles, pods and a live RACIOne owner for every kind of work
Operating cadenceRoadmap updates and direct notesStatus that reports itself
Executive reportingThe Operations Command CenterDecisions with context, not raw status
Delivery governanceA gated, seven-step release pathNo one checks their own work
Risk and controlsGuardrails and a compliance logClear lines a human must approve
Resource managementContext and cost monitoringCapacity treated as a budget

John P. Jackson

Co-Founder & Principal — The “human in the loop” at moonshinerXwerks LLC

John is a senior program manager with over 20 years of experience leading enterprise programs across PMO design, process optimization, commercial excellence, and global customer implementation. He spent a decade at Expedia's Egencia division, progressing from reporting and analytics through global implementation management to program leadership — overseeing Fortune 500 client delivery, consolidating regional teams into a unified global PMO, and earning a President's Club Award for onboarding $195M+ in annual new business revenue.

His career spans enterprise transformation and product-driven delivery at every scale: directing cloud migrations and SaaS platform rollouts, leading product development initiatives to improve SaaS usability and drive adoption, spearheading internal commercialization efforts that aligned product capabilities with go-to-market strategy, and managing Salesforce CRM integrations and commercial excellence programs across multiple organizations.

moonshinerXwerks is where all of that converges — an independent software studio applying 20+ years of enterprise discipline to consumer product development. Every app is taken from concept through release with full ownership of product strategy, feature delivery, privacy architecture, and commercial model. The full product lifecycle is managed using agile delivery, structured change analysis, and AI-enabled tooling — while also owning all supporting operations including legal compliance, hosting infrastructure, and CI/CD pipelines.

Based in Bend, Oregon. Usually found on a river when not managing the next product through development and delivery.

Want to get in touch?

Whether it's about one of our apps, a partnership idea, or just to say hello.

Contact Us →