2 slots left this month
The Lab · What I'm cooking

The systems I built to run the whole thing.

Agents, apps, and internal software that let an operator work with the leverage of a team ten times the size. The best ones graduate into products of their own.

130
systems in production
20
agents running daily
1
operating system
lab://status
Systems online130
Agents running20
Operating systemACTIVE
Last commit8 DAYS AGO
StatusSHIPPING
$ building next_
§ 01 — The ledger

The running total.

/ Updated October 2026
Total systems shipped●
130
Built, shipped, and running live today, not on a roadmap.
Skills●
90+
AI agents●
20
Funnels & sites●
8
Mobile apps●
6
SaaS products●
3
Automation tools●
2
The glue●
One operating system that ties every agent and app together.
§ 02 — Notable builds

Under the hood.

The systems I lean on most. Open any card for the reason it exists, not the feature list.

/ Seventeen of many

It's not one big assistant. It's a library of small, sharp operators, each built to do a single part of running the business well and consistently. Any process that gets repeated can be automated or streamlined.

What slows a solo operator down isn't the hard calls. It's the hundred small ones a day. These take the repeatable judgment off my plate so mine stays on what actually moves the number.

A single always-on system that holds live state, run history, approvals, and shared memory for the whole agent fleet. Agents push in, clients read out. The dashboard, Telegram briefings, and the CLI are all just interfaces into the same store.

Without this, every agent is an island. With it, the whole system knows what every other part just did.

Seven coordinated rails handling the full cycle: a daily decision engine, optimizer, reasoning layer, propose-only execution, creative pipeline, watchdog, and weekly intel report. It runs daily analysis, flags what needs attention, generates and queues creative, and produces its own calibration data so it gets sharper over time.

The grind of account management used to eat the mornings. Now the system handles the hundred daily micro-decisions and I focus on the strategy that actually moves spend.

Custom installs available for clients

Twelve coordinated skills that run the entire copywriting workflow: brand context, strategy, creative strategy, channel execution, copy review, and performance learning. Each skill owns one phase and hands off to the next.

This isn't 'ask AI to write a caption.' It's a production line where the research informs the strategy, the strategy constrains the creative, and the performance data feeds back into the next round.

Ten skills covering the full paid-growth loop: data ingestion, evidence research, funnel economics, funnel architecture, creative diagnosis, Meta performance analysis, and experiment prioritization. Each one produces a deliverable a strategist can act on.

Knowing which experiment to run next is the job. These turn raw ad data into ranked bets with the math already done.

Five review cadences (weekly, monthly, quarterly, annual, plus an opportunity filter) that pull session logs, daily notes, finance metrics, and ad performance before asking a single question. It interviews one question at a time and pushes back in the moment when an answer contradicts the data.

Its job is not to congratulate me for being productive. It's to determine whether I spent the period building the business I actually want.

One strong model plans the work and writes the acceptance tests before anything gets built. Cheaper and free open-weight models then build each task in its own isolated workspace, fenced to the files they're allowed to touch. Another model reviews every change cold, and nothing merges until I say yes.

Smart models are expensive and cheap ones are sloppy. This puts each where it's strong, so I get checked work without paying frontier prices for every line.

A self-hosted system that stitches the full journey, from first click to opt-in to purchase, and feeds real conversions straight back to Meta.

No tool on the market fit the workflow I use and the stack my clients run on, so I built my own.

It reads Substack reader evidence, builds a profile of the themes, generates product ideas from them, scans the market for what already exists, and positions the strongest ones. A full run takes a few minutes.

Most product ideas start as a guess about what people want. This starts from what they already said, in their own words.

A private email platform handling broadcasts, flows, segments, and tags, built on a shared identity core where a person is not a subscriber. The same Supabase instance powers attribution, funnels, and email, so every system sees the same contact timeline.

Off-the-shelf tools couldn't share a contact graph with attribution and funnels. Building my own meant one identity, one event stream, and no more duct-taping three platforms together.

Built around how an ADHD brain actually works. It holds what's in flight, what's next, and what I promised, then brings the right thing back at the right moment instead of asking me to keep it all in my head.

Most project tools assume you'll check them. This one does the remembering and the tracking, so I can just work.

A chat workspace for each client where I can think, draft, and decide, with their history, numbers, and notes already loaded. No re-explaining who they are every time I open a new conversation.

The context is the hard part of client work. Load it once and every conversation starts informed.

It produces the entire flow end to end: offer thesis, script, slide deck, and the hooks and funnel assets that get people to register and buy.

This turns an offer or idea into a finished sellable workshop in days, not weeks.

Spots the true outliers in organic, then builds the winner into paid creative, copy, and a campaign that's one click from live.

Most accounts guess at what to boost. This only spends behind what already earned its keep for free.

Coming soon to GitHub

Ideas flow constantly. They used to end up on a sticky note that got lost, or they were forgotten entirely. Now the moment one shows up, I drop it in and forget about it. It clusters recurring ideas and surfaces them at the exact moment I need them, so I can plan my day strategically and keep my head free for real thinking.

Notes apps solved capture. They never solved remembering to look. This one does the remembering for you.

A unified agent that triages email, newsletter replies, comments, and social media DMs into a single queue.

Nothing important slips through. Nothing unimportant steals my attention.

It scores an existing workshop or webinar not just on what it can write, but on what actually converts and how you're positioned against your competitors, then hands back a scorecard showing exactly where it's leaking.

Built around your positioning, not a generic template.

Not just for me

What I'm building for clients.

Most of this I built for myself first. A handful of operators have me build the same kind of leverage into their businesses, quietly, behind the scenes.

/ Private engagements
№ 01

Custom dashboards

One screen for the numbers that actually run the business. No more manually filling in spreadsheets or tab-hopping to see how it's doing.

№ 02

An AI Chief Marketing Officer

An agent installed to run the day-to-day of your marketing, on brief and on brand, reporting to you.

№ 03

Custom workflows

The repetitive parts of your operation wired to run themselves, end to end.

№ 04

Custom skills

Narrow AI tools built for the exact jobs your team repeats every week.

№ 05

Funnel & lead-gen systems

The acquisition machine, built, wired to your stack, and handed over working.

№ 06

…and more

If it's repeatable and it's slowing you down, it can probably be built. That's the whole game.

You might not need to hire another team member. You might just need a system or workflow built to do the job that's currently missing.

On why the tools are mine · Victoria, BC

I build software, AI systems, workflows, products and operating systems that make business dramatically simpler. It starts with mine.

Every system on this page started as a problem I'd stare at on a Tuesday morning. These translated into hours of repetitive work, tasks or projects I'd never complete, or thousands of dollars spent (attempting) to solve.

This isn't about using ChatGPT to write an Instagram caption.

My goal is leverage. More importantly, it's in focus. Every system amplifies what I do or removes something from my plate so my focus can be on my highest value and enjoyable work. Oddly, that's usually on solving more problems ;-)

After two decades, what I love doing and what I do for work have united.

— Landon
§ 03 — The stack

What it's built with.

/ Updated 2026
Editor
Claude Code + VS Code
Hosting
Vercel
Database
Supabase
Design
Claude Design
Notes
Obsidian
Voice
Wispr
Network
Tailscale
Security
1Password
Meetings
Granola
Version control
GitHub
Model routing
OpenRouter
Open models
Gemma · Qwen · DeepSeek
Harness
OpenCode
Next up
Android devices
Next up · Hardware

Teaching physical devices to talk to the stack.

First build: an always-on Android panel on my desk streaming the whole system live, a mission-control readout for the business. Software was the warm-up; hardware is where it gets fun.

Build with me

These are mine.
Yours could be next.

Every system here, I built to run my own business first. If you want one like it running yours, let's talk through what that looks like.

Book a strategy call →
2 slots left this month
Response within 48 hours · Victoria, BC