📅 Sunday weekly review — pending since 2026-06-21
Next week is unplanned until you run it. ~50-65 min, ends with Monday's first move + deep-work block committed.
/weekly-review
Monday, June 22
Today.
Sunday committed this week’s focus. Execute the move. Override only with written tax.
This week — magier MIT
Claude Code workshop for art directorsPrep + deliver a session teaching the team basic Claude Code concepts
Today’s first move
Draft the workshop outline — list the basic Claude Code concepts to teach the art directors, in the order you'll teach them. ≤30 min, fits between Monday's meetings. (Parallel small task for any gap:…
MITs
magier
Claude Code workshop for art directorsPrep + deliver a session teaching the team basic Claude Code concepts
LeadBridge Local
Close last month's LB accountingCollect the invoices + wrap the books, then LB goes dormant to 2026-09-01
⏱ accounting closed during W25,
TrueToYou
No MIT this week — tend mode
Erbright
Launch the Erbright websiteFinish the ~80%-built proof site (separate build chat) + ship it live, minimal branding
⏱ none hard yet (contained one-w
The week
Mon · today 22
9:00🧙 Weekly Check-In
Tue 23
9:20Flight to Rome (Fiumicino) (FR 40)
10:00OTP - Pulse Check
14:00Commerzbank anrufen wegen Kontoauszügen Nachlass
Wed 24
10:00AI Design Gang 😎🪄
15:45Ojas x Willi | Bi-Weekly
Thu 25
14:00magier - Weekly
14:30Magier x Webflow - Co-Marketing sync
16:00Alessa x Willi | Q 1:1
16:30Mo x Willi | Q 1:1
18:00Willi x Max (1:1)
Fri 26
9:00🧙 Weekly Check-In
11:00Oma Gisela
13:00magier - OKR - [Recap + Sailing Boat]
Sat 27
no events
Sun 28
no events
Override log
No overrides logged. Sunday-Willi’s plan is holding.
OS meta
System.
The machine looking at itself — audit score, skills, memory, wiki, decisions. Reality of the OS, not the day.
Data health
4 aging●Calendar1d old
●Email / inbox digestAI-sort degraded — emails fetched, triage fell back (often a transient model blip)
●Granola sync3d old
●Job: morning-briefstale — 40h old (budget 30h)
●Dashboard deploy (droplet)10m ago
●OKRs7d old
●OS audit10d old
●Revenue / bonus7d old
●magier dream-ideas5d old
●magier insights5d old
OS scorecard
as of 2026-06-13 (9d ago)Context
12/25
Manual + decision log are strong; no voice-core, no narrative layer, MEMORY.md over its size limit, wiki is a 1-source skeleton
Connections
16/25
All 7 Tier-1 domains reachable, write breadth good; reference docs and per-wire freshness dates mostly missing
Capabilities
17/25
7+ skills used weekly with real data; dead-skill clutter and the single-agent gate cost points
Cadence
22/25
9 live launchd jobs, daily auto-artifacts, event hooks, no zombies — the strongest layer
Next moves to raise it
1
Build the Context narrative + voice-core layer
Effort: M · Payoff: daily · **Why first:** it's the lowest-scoring layer, it's the Constitution's own pending Session 2, and it unblocks every voice-dependent s…
2
Trim MEMORY.md back under the 24.4KB limit
Effort: S · Payoff: daily · **Why:** it's an active failure right now — the index only partially loads every session, so recall silently degrades. Move long ind…
3
Drift pass on OS.md + connections.md
Effort: S · Payoff: weekly · **Why:** the map is 3 weeks stale and audit accuracy depends on it. Update the Cadence section to the real 9 launchd jobs, strip de…
OS health
OS audit
67/100
Built
Skills
77
global
Memory files
161
atomic
Wiki sources
1
ingested
Inbox
0
awaiting Sunday
Overrides
0
this period
Inbox
Inbox empty. Sparks land here via
/idea <line>. Cleared every Sunday review.Skill library
74 skills · type the trigger in chatPersonal / OS
Day-to-day Joy OS — capture, plan, review, meta. · 25
/audit-os
audit-os
Looks at how healthy your whole personal operating system is right now. It scores four layers — context, connections, capabilities, and cadence — then names the weakest one and tells you the top few things to fix first.
/close-bet
close-bet
The companion to idea validation. When a bet you set up earlier hits its end date, this judges whether it actually cleared the threshold you committed to. The verdict is one of three: go bigger, kill it, or iterate.
/cross-research
cross-research
Runs two different research engines on the same question in parallel, then shows you where they disagree. Used when you want to compare two independent takes before making a decision that actually matters.
/dashboard
dashboard
Rebuilds your dashboard from the current state of everything and opens it. Use it when you've made changes and want a fresh view of where things stand right now.
/dream
dream
This skill helps you get new ideas each day by looking through your recent notes, conversations, and important topics. It gives you a few simple idea cards with a quick summary and a suggestion for how you might use them. You'd use it when you want fresh inspiration or to brainstorm something new.
/i-have-adhd
i-have-adhd
Switches the way I write to you for the rest of the session. Instead of long paragraphs, you get numbered steps, time estimates, and no tangents. Useful when you just need to execute and the bigger context is getting in the way.
/idea
idea
For when a thought hits you mid-work and you don't want to lose it. It captures the spark in under ten seconds, then asks three quick questions while it's still fresh. Sunday's review picks it back up from there.
/intel-weekly
intel-weekly
Runs the whole weekly intel pipeline in one shot. It pulls fresh AI, coding, and design videos from YouTube, digests around ten of them, and writes a ranked weekly digest. Heavy on quota, so most weeks the split version is safer.
/level-up
level-up
Helps you find the next thing worth automating. It asks five questions about where time is leaking, looks at which of your existing skills are actually getting used, and recommends one concrete next move you can build this week.
/morning-brief
morning-brief
Your morning kickoff. It tells you what lane you're in today, the one thing that matters most, and the very first move to make. Pulls from the plan you set on Sunday so you don't have to think about it before coffee.
/refine-skill
refine-skill
When a skill almost did what you wanted but missed something specific. This makes a surgical edit to its instructions so it does better next time, without rewriting the whole thing.
/safe-schedule
safe-schedule
Before you set something up to run on a schedule — a daily job, a weekly digest, a reminder — this forces a one-minute test run first. The whole point is to catch broken schedules before they fail silently for weeks while you're not looking.
/session-handoff
session-handoff
When you want to close out a session cleanly. It writes a short summary covering what got decided, what shipped, and what's still open, so a fresh session can pick up exactly where you left off with all the context it needs.
/skill-builder
skill-builder
The fast way to make a new skill. It asks five questions, writes the skill file, and registers the slash command. You go from a vague want to a working skill in a few minutes.
/skill-optimizer
skill-optimizer
Trims a skill's instructions down to use fewer words while keeping it working exactly the same. Useful when a skill has slowly grown bloated over many edits.
/skill-usage
skill-usage
Reads your session history and counts how often each of your skills actually got used. Helps you see which ones earn their keep and which are dead weight that could be retired.
/task
task
When you've already decided you need to do something concrete, this drops it under the right initiative in the right lane. Different from idea-capture: this is for work that's committed, not for thoughts you still want to chew on.
/three-brain
three-brain
For tough work, this routes the task to whichever model is best suited — Claude, Codex, or Gemini. Especially important when you want a second opinion on something I just wrote, because I can't fairly review my own output.
/validate-idea
validate-idea
Turns a new business idea into a binding bet with a clear pass-or-fail line that you commit to before you start. It's not for brainstorming. It's for the moment you've already had the idea and want to test it properly before you sink real time into building.
/weekly-review
weekly-review
The Sunday session that closes one week and sets up the next. It walks you through clearing your inbox, picking next week's most important thing, and putting your big deep-work block on the calendar. Takes around an hour and is the most important ritual of the week.
/wiki-ingest
wiki-ingest
Saves an article, transcript, or PDF into your personal wiki as readable notes. The notes get cross-linked to anything related you've already saved, so over time your knowledge connects itself together.
/wiki-query
wiki-query
Before reaching out to the open web, this asks your own wiki first. Useful when you remember reading something somewhere but can't recall the source, or when you want an answer grounded in what you've already curated.
/evening-diary
evening-diary
A short wind-down at the end of the day. You reflect on how things went and capture what you learned before sleep, so the lesson doesn't evaporate overnight.
/intel-collect
intel-collect
The mid-week version of intel. It runs Monday, Wednesday, and Friday — finding fresh candidate videos and digesting about ten of them per run. Stays inside the free daily quota so you can run it all year without thinking about it.
/intel-compile
intel-compile
The Sunday step that closes the intel week. It ranks everything you collected during the week and writes the final weekly digest you can actually read.
LeadBridge Local
Rank-and-rent pipeline + portfolio ops. · 18
/analyze-niche
analyze-niche
Screens a local service niche to see if it's worth building a site for. It looks at how many people search for the service, what advertisers pay per click, and what the Google results page looks like. Gives you a fast go or no-go on the first pass.
/audit-portfolio
audit-portfolio
A monthly check across every live LeadBridge site at once. It generates a snapshot of how each site is performing and a prioritized list of what to fix where. Tells you where your effort should go next month.
/build-niche
build-niche
The full pipeline in one command. It goes from a niche idea all the way to a deployable site by running every step in order. Useful when you've already validated the niche somewhere else and just want to ship without thinking.
/build-site
build-site
Takes the finished German copy and turns it into actual HTML pages using the LeadBridge template. This is the last step before you can deploy a site live.
/compress-images-avif
compress-images-avif
Converts the images on a site from regular formats like JPG to AVIF, which loads way faster in a browser. It moves the originals into a backup folder and updates every place the images are referenced.
/deep-analysis
deep-analysis
After a niche passes the first screen, this digs deeper into the competition and their backlinks. You paste the backlink data in manually because the automatic source isn't reliable enough. The output is a final verdict on how hard it would actually be to rank.
/find-backlinks
find-backlinks
Finds backlink opportunities for a site by looking at who's already linking to the top-ranking competitors. It then tells you which of those links you can realistically replicate yourself.
/generate-citations
generate-citations
Builds a tailored list of business directories and citation sites where you should submit a new site. Different trades and cities need different directories, and this picks the right ones automatically.
/local-seo-content-planner
local-seo-content-planner
Plans the structure of a new site — which pages it should have, which keywords go on which page, and how they all link together. Runs after the niche is approved but before any copy is written.
/local-seo-copywriter
local-seo-copywriter
Writes the actual German copy for every page on a site. It reads from the content plan so each page targets the right keywords without you needing to keep track of which keywords belong where.
/local-seo-image-pipeline
local-seo-image-pipeline
This skill automatically creates and perfects images for websites using AI. It makes sure the pictures look great, are the right size, and are ready to be used online. You'd use it whenever you need to get eye-catching images ready for your website quickly and easily.
/local-seo-image-prompts
local-seo-image-prompts
Generates the AI image prompts for every visual slot on a site — hero shots, service photos, team photos, before-and-afters. You paste them into Freepik and get the images back.
/preflight-build
preflight-build
Locks in the boring-but-critical decisions before building — the domain name, the brand voice, what's in scope and what isn't. It runs after the writing and image phases are done, but before you start assembling the actual site.
/prioritize
prioritize
Walks you through your LeadBridge dashboard so you can decide what to work on next. I propose a re-ranking based on what I see, but you always have the final call.
/qa
qa
Checks a built site before you deploy it. It looks for leftover placeholder text, SEO problems, broken links, bad structured data — the whole list. Always run this before pushing a site live.
/qa-live
qa-live
The check that runs after a site is already live. It verifies DNS, HTTPS, Lighthouse scores, and that the canonical URL is set up right. Run it after every deploy.
/quarter-review
quarter-review
The end-of-quarter ritual for LeadBridge. It looks at what shipped versus what slipped, checks whether you're still aligned with your yearly outcome, and commits three to five concrete bets for the next quarter.
/site-tasks
site-tasks
For managing the small to-do list on an individual LeadBridge site. You can add tasks, check them off, or remove old ones, and each site has its own clean backlog.
Cross-cutting tools
Research, web, content, design — usable anywhere. · 31
/brand-capture
brand-capture
This tool lets you grab all the design details like colors, fonts, and styles from any website and save them. It creates a special collection of these design rules so you can easily use them again whenever you need to.
/deep-research
deep-research
The heavy machinery for research that actually matters. It spins up seven research agents in parallel, has them double-check each other, and synthesizes one final report. Don't use it for quick lookups — it's expensive and slow, but rigorous when the decision is real.
/design-language
design-language
This skill helps you understand the unique visual style of any website you like. It breaks down what makes that site look cool—things like colors, fonts, and overall layout—so you can use those ideas to create your own projects with a similar feel.
/design-system-creator
design-system-creator
Builds a complete design system from scratch — color tokens, spacing, type scale, a component library, and the docs that go with them. Made for greenfield design work, not patching styles on an existing site.
/firecrawl
firecrawl
The general web-access tool. You can search the web, scrape pages, or click around on dynamic sites, all from one place. Use this when you're not sure which specialized variant you need.
/firecrawl-agent
firecrawl-agent
For when you want structured data from a multi-page site. You give it a shape to fill — like pricing tiers, product listings, or directory entries — and it navigates the site filling that shape in automatically.
/firecrawl-build-interact
firecrawl-build-interact
Wires browser interaction into product code. Use this when your app needs to click, fill forms, or log in to reach the content it needs.
/firecrawl-build-onboarding
firecrawl-build-onboarding
For when you're adding Firecrawl into an app you're building. It sets up the API key, picks the right SDK, and gets the first integration running so the rest of the wiring goes smoothly.
/firecrawl-build-scrape
firecrawl-build-scrape
Wires single-page scraping into product code. Use this when you're writing an app that needs to grab a webpage on demand and return its content.
/firecrawl-build-search
firecrawl-build-search
Wires web search into product code. Use this when your app needs to discover URLs from a query before it can extract anything from them.
/firecrawl-crawl
firecrawl-crawl
Pulls every page out of a website or a section of one. Good for grabbing all of a docs site or every listing under a category in one go.
/firecrawl-download
firecrawl-download
Saves an entire website to your computer as local files. Useful for offline reference or making your own copy of documentation you rely on.
/firecrawl-interact
firecrawl-interact
Drives a live browser session — clicking buttons, filling forms, paginating, logging in. Use it when plain scraping fails because the content you want lives behind clicks or logins.
/firecrawl-map
firecrawl-map
Lists every URL on a website, with an optional keyword filter. Use it when you know the site but don't know which exact page you're looking for.
/firecrawl-scrape
firecrawl-scrape
Grabs a single webpage and returns it as clean, readable text. It handles JavaScript-heavy sites that normal scrapers usually choke on.
/firecrawl-search
firecrawl-search
Searches the web and returns the full content of every result, not just snippets. More useful than normal search when you want to actually read what came back, not just preview it.
/graphify
graphify
Takes anything — code, documents, papers, images — and turns it into a knowledge graph you can ask questions of. Useful when you have a big pile of stuff and want to query across all of it at once instead of opening files one by one.
/linkedin-content-magic
linkedin-content-magic
Drafts a LinkedIn post in your voice across four post types — personal stories, news takes, education, and infographics. It tracks the balance so you don't accidentally post the same type three times in a row.
/notebooklm
notebooklm
Full programmatic access to Google's NotebookLM. You can create notebooks, add sources, generate podcasts and briefings, and download everything. Goes beyond what the web interface lets you do.
/printing-press
printing-press
This skill helps you quickly make a command-line program that can connect and talk to other computer systems. It's like having a fast conveyor belt that takes your idea from planning to a finished, ready-to-use program in no time.
/printing-press-amend
printing-press-amend
This skill lets you fix or update content after it's been made, like correcting a mistake or adding new details to something that's already published. You'd use it whenever you need to adjust information that's largely complete.
/printing-press-catalog
printing-press-catalog
This skill helps you discover and set up ready-to-use computer programs that you control by typing commands. These programs let you easily connect and work with popular online services. It's like having a special shop where you can pick out and install tools to do cool things on the internet.
/printing-press-import
printing-press-import
I need the original description for the "Printing Press Import" skill. Please provide it so I can rewrite it as requested.
/printing-press-output-review
printing-press-output-review
I need the original skill description to rewrite it. Please provide the "Original description" content.
/printing-press-polish
printing-press-polish
This skill makes your writing and other content look super good and professional, like it's been perfectly prepared for publishing. It's great when you want to give your work a final, impressive finish before sharing it.
/printing-press-publish
printing-press-publish
This skill helps you share a new command-line program you've made, similar to putting a new app in an app store. You'd use it when you want others to be able to download and use your program from a central collection.
/printing-press-reprint
printing-press-reprint
I cannot rewrite the skill description as the 'Original description' field is empty.
/printing-press-retro
printing-press-retro
This skill helps you make your text and documents look like they were made on an old printing press. You can use it to give your school projects or personal notes a cool, vintage style.
/printing-press-score
printing-press-score
This skill helps you check how well a command-line program works compared to a high standard. You can also use it to compare two different command-line programs side-by-side.
/send-letter
send-letter
This skill helps you send a real letter in the mail to anyone, like a court, a company, or just a friend. It's super useful when you're traveling and can't print or mail things yourself.
/vibe-watch
vibe-watch
This skill helps me find and understand the special notes you've added to a webpage. You'd use it when you want me to look at those notes and act on the things you've marked on a site.
OS initiatives
13 open · cross-cuttingJoy OS v1 — lane→initiative→task hierarchy
Activeclean separation between *ideas* and *tasks* ; makes weekday execution zero-decision.
5/7 done
Pattern-mining across calls + Claude chats
Active/granola-digest already mines Granola call transcripts into actionable dream-ideas . Willi wants the same pattern recognition applied to *Claude chat sessions* — recurring requests…
0/6 done
Productivity OS hardening — post-v1
ActiveOS Constitution + audit infra ratified 2026-05-22. Two-week post-build review scheduled per design.
2/16 done
Steal-from-gstack — port pressure, not names
Activegstack is the most thought-through public Claude Code stack. Six patterns + the artifact-chaining meta-pattern map cleanly to gaps in Willi's OS.
0/9 done
Map & steal from Jack Roberts' Claude setup
ActiveWilli bought into Jack Roberts' paid community . Sibling to the gstack port but a different source — Jack's stack is the thing being mapped here.
0/1 done
Obsidian as substrate over ~/Claude
ActiveConvert `~/Claude/` into an Obsidian vault — human UI layer over the existing markdown OS. Doesn't change runtime; unlocks graph view, backlinks, mobile reading, vault portability.…
0/7 done
Always-on cadence engine on Hetzner VPS
ActiveCadence engine must fire regardless of laptop state — going fully nomadic 2026, deregistering from Germany, no permanent home for a Mac mini.
0/13 done
Magier Executive Assistant build queue
ActiveMagier is 90% of cognitive load; the EA-build queue is the systematic path to remove Willi from the day-to-day.
0/8 done
End-state EA surfaces — DEFERRED parking lot
PausedThree EA end-state visions that depend on the EA build queue producing data first. Captured to remember the destination; not buildable until the feeders exist.
0/3 done
M008 morning-brief refactor — code-over-prompt
ActiveFirst concrete application of M008 . Extract `/morning-brief` to `morning_brief.py` , slim skill to `/override-plan`, then add Override button with Web Speech API → 3 sequential vo…
0/3 done
iPhone mic — ambient day-recording into Claude context
ActiveHighest-level context Claude could ever have — keeps memories continuously up to date without manual capture. From W21 inbox 2026-05-23.
0/6 done
Print `webflow-pp-cli` via Printing Press
Pausedreplaces the broken Webflow Designer API + official MCP for magier's workflow. Unlocks Figma → Webflow pixel-perfect conversion without Designer being open, and Slack → Webflow pag…
4/10 done
Foundation Sprint skill (`/foundation-sprint`)
ActiveJake Knapp's Foundation Sprint is the upstream clarity/positioning piece the current `/validate-idea` → `/close-bet` pipeline is missing.
0/2 done
Webflow & graphic design agency
magier.
Profit
€99.106 / €1.000.000
Margin
9.3% / 30%
€7.929
Bonus YTD
8% tier · proj. €19.028
OKRs — Q2 2026
from Notion · 4 key resultsOTP NPS
94%
85 / 90
Implementing our first 3 internal agents or tools
100%
3 / 3
New OTP Structure creates healthy margins
95%
62 / 65
magier Studios booked new €250k OTP Revenue
61%
319k / 522k
Dream ideas
Bot-blocked sitemap fallback with CSV intake
Pain pattern
When a client host blocks bots, the migration workbench returns an unhelpful error and the team has no documented fallback, forcing manual workarounds that vary by person. · Seen in: 2026-06-12-0800-migration-workbench-intro-83Ga.md
Suggestion
A fly.io worker or n8n flow that detects a bot-block response during sitemap fetch, triggers a structured CSV intake form sent to the client via email, parses the returned CSV into the workbench schema, and posts a Slack notification to the project channel when the import is ready for review.
Implementation hint
Extend the migration workbench's existing sitemap fetch logic with a bot-block detection branch. On detection, the tool renders an in-app banner with step-by-step instructions and triggers an n8n flow that emails the client a pre-filled CSV template with the required columns: page name, URL slug, title tag, meta description. When the client returns the CSV, the n8n flow parses it, validates the schema, and pushes the rows into the workbench via its existing API. The Ariba project is the first live test case and the CSV template columns are already known from the session.
Why now
Every enterprise client with bot-blocking hits this gap on the first sitemap fetch, and without a documented path the team improvises differently each time, creating inconsistent project starts that affect SEO handoff quality.
Prospect warm-handoff brief generator for incoming decision-makers
Pain pattern
When a prospect's key decision-maker is not yet in role, the window between first contact and their onboarding closes before magier can get a tailored pitch in front of them. · Seen in: 2026-06-15-0800-sync-moritz-willy-mustafa-Ptb3.md
Suggestion
A Claude agent that accepts a call transcript and a prospect company URL, extracts the stated pain points and tech stack, and generates a one-page PDF brief tailored to the incoming decision-maker's domain, including relevant case studies, a positioning paragraph, and a suggested first-meeting agenda.
Implementation hint
Build a /prospect-brief skill that takes a Granola transcript basename and a company URL as inputs. Claude extracts pain points, tech mentions, and decision-maker domain from the transcript, then firecrawl-searches the company site for additional context. The output is a structured markdown brief with four sections: their situation as heard, relevant magier case studies matched by industry and tech stack, a positioning paragraph for the decision-maker's domain, and a suggested agenda for the follow-up call. The brief is rendered as a PDF via a fly.io worker and posted to the relevant Slack sales channel. The Vays case is the first pilot: Niklas is CMO-domain, so the brief should lead with AI-augmented marketing and web, not design subscription.
Why now
The Vays deal is warm but structurally paused until Niklas signs, and the portfolio pack Willi committed to sending needs to land before the notary date, not after, or the first impression defaults to whatever the other pitches say.
Ebook style frame generator from brand profile
Pain pattern
Every new ebook or editorial deliverable requires a style definition round from scratch because magier has no reusable editorial layout system, consuming the buffer before copy even arrives. · Seen in: 2026-06-15-1202-weekly-telli-x-magier-FDOA.md, 2026-06-17-1259-george-x-willi-wbhs.md
Suggestion
A Claude agent that accepts a client brand profile JSON and a content type tag such as ebook or one-pager, generates two Figma-ready style frame specifications covering typography scale, grid, color application, and section templates, and posts them as a structured brief to the relevant Many Requests ticket before copy is received.
Implementation hint
Build a /editorial-style-frame skill that reads the client brand profile JSON from the clients folder and a content-type parameter. Claude generates two style direction specs: one conservative applying the existing brand system directly, one more expressive pushing one brand element such as color or typography. Each spec includes: type scale, grid columns, primary and accent color usage rules, and three section templates described in enough detail for a designer to build in Figma without a reference call. The output posts as an internal ticket note in Many Requests. The Telli ebook is the first pilot since copy is arriving this week and the style gap is already blocking parallel work. The brand profile JSON infrastructure being built for the brand guidelines agent provides the shared input layer.
Why now
Telli has a hard Tuesday June 23 deadline for a 30-page ebook with no style precedent, and every day spent on style definition after copy arrives is a day removed from design execution time.
Shadow IT policy auto-enforcer for Claude connectors
Pain pattern
Team members are granting Claude connectors access to client Figma files and live websites before a data policy exists, creating untracked exposure that varies by individual judgment. · Seen in: 2026-06-17-0800-claude-basics-intro-eS0D.md
Suggestion
A Claude agent that runs as a weekly audit, reads the list of active Claude connector permissions across the team, flags any connector granted to a client-tagged project without a corresponding approved-use entry in the shadow IT rulebook, and posts a summary to a designated Slack channel for Willi and Gabriel to review.
Implementation hint
Build a /connector-audit skill that reads a team-level connector registry maintained as a markdown file in the magier Claude folder, cross-references each entry against an approved-use list in the shadow IT rulebook, and generates a weekly diff showing new connectors added since the last audit, their granted scope, and whether they have rulebook coverage. The audit runs as a scheduled task in the Claude desktop app every Monday morning and posts to a private Slack channel. The approved-use list starts with the four interim rules Willi stated in the workshop and expands as the rulebook is built. The first run uses the connectors Alessa and Vincent already have active as the baseline.
Why now
The team is already using Claude on client work without guardrails and the rulebook does not exist yet, meaning every week without the audit is a week of untracked exposure that grows as more team members activate connectors after the workshop.
OTP blocker tracker with client-comms nudge
Pain pattern
Multiple OTP projects stall for days waiting on client-side inputs such as copy approval, CMS content, or partner section details, with no automated follow-up and no visibility into how long each project has been blocked. · Seen in: 2026-06-16-0801-otp-pulse-check-jP0Y.md, 2026-06-12-0659-weekly-check-in-2QpM.md
Suggestion
An n8n workflow triggered by a Many Requests ticket status change to On Hold that starts a countdown, and if the ticket remains On Hold for more than 48 hours without a client reply, automatically sends a templated follow-up message to the client via the ticket activity feed and posts a Slack alert to the project manager.
Implementation hint
Build an n8n flow triggered by the Many Requests webhook on status change to On Hold. The flow records the timestamp and sets a 48-hour timer. If no client activity is detected in the ticket within 48 hours, Claude drafts a follow-up message using the ticket title, the last PM comment, and a one-sentence prompt asking the client for the specific missing item. The draft posts as a pending message in the ticket activity feed for PM review before sending, not automatically to the client. A Slack notification goes to the assigned PM with a one-click approve-and-send button. JetBlend, DeliWell, and Lead and Security are all currently in this state and would have triggered this flow if it existed.
Why now
JetBlend was blocked for a week on clarification questions that had to route through Sina before reaching the client, and DeliWell is stalled on a script with slow client comms, meaning the cost of each day of delay is direct margin erosion on fixed-scope projects.
Initiatives
1 openCustomer & Former-Customer Interviews (ICP Discovery)
ActiveLearn who magier's real ICP is and what pain we actually solve, via Mom-Test-style interviews with current + churned customers.
0/8 done
Rank-and-rent SEO sites
LeadBridge Local.
€5k/month recurring by Dec 2026.
Initiatives
7 open · your task boardMonthly accounting
Activerecurring monthly bookkeeping for LeadBridge. Willi 2026-06-08: collect last month's invoices + wrap the accounting this week.
0/2 done
Phone number replacement after EasyBell cancellation
ActiveEasyBell is cancelling all four LeadBridge geographic numbers 2026-05-28 per BNetzA Allgemeinverfügung 25/2006 — a US LLC cannot legally hold Ortsnetzrufnummern.
0/6 done
Novel infrastructure / new skills
Activerolls up `/level-up` candidates as they emerge from Sunday review.
0/4 done
AI-agent readiness — llms.txt + Bing WT + agentic-browsing audit
ActiveGoogle IO 2026 shipped browser search agents + personalized dashboards. Click-driven SEO is decaying — measured KPIs shift to citations + search appearances.
0/10 done
District-level content for agent recommendation
parkedSame Google IO 2026 thesis as the llms.txt initiative — agents know user context even when the user doesn't type it.
0/8 done
LB-improve backlog — PARKED candidates
parkedsparks that are good but not the right thing to build now.
0/3 done
Backlog — deferred LB work
Activesee `~/Claude/leadbridge-local/BACKLOG.md` for the canonical list. Top item: EasyBell voicemail-to-email setup. This section mirrors only what `/task` adds via the markdown layer.
0/1 done
Pipeline
committed to build → rentedMRR
€0
target €5k/mo
Live
15
deployed
In build
0
active
Pipeline
0
committed, queued
49 analyses ready for your build / skip decision — see “Ready to build” below. Not counted as pipeline.
0Build queue
→0Building
→15Live
→0Rented
Needs you · 0
Nothing needs you right now.
⚠ 8 niches are missing an expected field for their stage (volume / CPC / bootstrap / ref-domains). Their cells show “—” — the data exists to be filled by re-running the relevant analysis, it is not a real zero.
Live sites · 15
In build & blocked · 11
Building (0)
none
Blocked (11)
alarmanlage-potsdamalarmanlage-rostockdachrinnenreinigung-hamburgestrich-potsdamestrich-rostockestrich-ulmfliesenleger-rostockinnenausbau-berlininnenausbau-hamburgpraxisreinigung-berlintrockenbau-berlin
Ready to build · 49
Deep-analyzed. Favour high bootstrap + high cluster vol × CPC + low referring-domain counts (weak competitors).
| Niche | Bootstrap | Cluster vol | CPC € | Top-5 ref domains |
|---|---|---|---|---|
| entruempelung-rostock | 8.6 | 400 | 4.97 | 3, 5, 109, 49, 54 |
| abbrucharbeiten-berlin | 7.8 | — | — | 480, 19, 114, 61, 79 |
| elektriker-notdienst-hamburg | 7.7 | 290 | 7.53 | 48, 87, 328, 38 |
| abbrucharbeiten-hamburg | 7.5 | — | — | 207, 149, 53, 72, 54 |
| abbrucharbeiten-muenchen | 7.5 | — | — | 42, 180, 132, 78, 72 |
| elektriker-notdienst-berlin | 7.5 | 1050 | 5.66 | 72, 85, 63, 227, 111 |
| elektriker-rostock | 7.5 | 640 | 3.32 | 49, 53, 66, 64, 59 |
| fensterbauer-ulm | 7.5 | 1560 | 2.92 | 81, 103, 120, 3, 101 |
| tiefbau-berlin | 7.4 | 2270 | 9.41 | 176, 100, 84, 68, 2 |
| elektriker-notdienst-muenchen | 7.2 | 210 | 7.63 | 47, 49, 224, 44 |
| entkernung-berlin | 7.2 | — | — | 44, 480, 193, 19 |
| estrich-berlin | 7.2 | 570 | 3.78 | 63, 51, 67, 331, 79 |
| fensterbauer-rostock | 7.2 | 290 | 3.59 | 131, 67, 78, 54, 149 |
| fliesenleger-ulm | 7.2 | 780 | 2.8 | 63, 52, 110, 93, 163 |
| kellerabdichtung-hamburg | 7.2 | 270 | 7.77 | 160, 160, 60, 31, 30 |
| estrich-hamburg | 7.1 | 280 | 4.62 | 69, 84, 331, 19, 5 |
| gartenbau-rostock | 7.1 | 1230 | 2.23 | 108, 74, 138, 77, 45 |
| badsanierung-potsdam | 7 | 1720 | 9.18 | 142, 374, 33 |
| estrich-muenchen | 6.9 | 390 | 2.84 | 107, 39, 31, 42, 286 |
| fliesenleger-potsdam | 6.9 | 470 | 3.61 | 81, 1300, 54, 54, 56 |
| dachdecker-potsdam | 6.8 | 440 | 2.32 | 115, 69, 80, 146, 60 |
| fensterbauer-muenchen | 6.8 | 2420 | 5.62 | 203, 219, 209, 48, 116 |
| gartenbau-ulm | 6.8 | 1730 | 2.13 | 54, 4, 38, 45, 45 |
| tiefbau-muenchen | 6.8 | 1400 | 4.84 | 130, 82, 71, 20, 14 |
| innenausbau-muenchen | 6.6 | 210 | 4.4 | 83, 315, 212, 51, 35 |
| tiefbau-rostock | 6.6 | 310 | 1.62 | 86, 19, 94, 77, 45 |
| fensterbauer-potsdam | 6.5 | 260 | 2.23 | 19, 111, 134, 98, 87 |
| dachdecker-ulm | 6.2 | 530 | 2.29 | 107, 79, 168, 68, 140 |
| wasserschaden-bautrocknung-potsdam | 6 | 170 | 7.57 | 86, 115, 62, 85, 38 |
| alarmanlage-berlin | 5.8 | 1290 | 13.53 | 252, 153, 233, 228, 130 |
| gartenbau-potsdam | 5.8 | 950 | 2.25 | 8, 85, 69, 141, 52 |
| tiefbau-hamburg | 5.8 | 1080 | 3.79 | 360, 286, 89, 39, 19 |
| alarmanlage-ulm | 5.7 | 60 | 5.17 | 112, 237, 146, 58, 88 |
| kellerabdichtung-berlin | 5.7 | 290 | 24.08 | 94, 190, 89, 91, 76 |
| entruempelung-hamburg | 5.5 | 7210 | 10.22 | 117, 142, 149, 230, 128 |
| entruempelung-muenchen | 5.5 | 5130 | 7.43 | 177, 133, 85, 251, 212 |
| alarmanlage-muenchen | 5.2 | 380 | 5.49 | 124, 162, 354, 161, 131 |
| gartenbau-berlin | 5.1 | 6580 | 3.24 | 183, 234, 128, 21, 26 |
| elektriker-potsdam | 5 | 600 | 3.57 | 74, 75, 49, 147, 29 |
| entruempelung-berlin | 5 | 12660 | 8.12 | 137, 272, 277, 145, 152 |
| fassadenreinigung-hamburg | 5 | 320 | 29.79 | 106, 56, 102 |
| fensterbauer-berlin | 5 | 3460 | 5.74 | 249, 314, 66, 177, 181 |
| treppenlift-berlin | 5 | 260 | 16.88 | 458, 66, 56, 48, 47 |
| treppenlift-muenchen | 5 | 280 | 20.95 | 231, 139, 97, 89, 65 |
| tiefbau-potsdam | 4.9 | 170 | 4.72 | 92, 234, 48, 162, 71 |
| treppenlift-hamburg | 4.8 | 200 | 18.95 | 240, 186, 126, 71, 53 |
| alarmanlage-hamburg | 4 | 480 | 10.95 | 403, 146, 130, 566, 126 |
| fensterbauer-hamburg | 4 | 3110 | 5.78 | 158, 374, 32, 174, 594 |
| schimmelbeseitigung-ulm | 4 | 200 | 5.95 | 2300, 135, 117 |
Ready to deep-analyze · 33
Demand-screened only — bootstrap + competitor backlinks appear after /deep-analysis. Prioritise high cluster vol × CPC.
| Niche | Cluster vol | CPC € | avg KD |
|---|---|---|---|
| elektriker-berlin | 6940 | 3.95 | 22.2 |
| fliesenleger-berlin | 6890 | 3.77 | 36 |
| buero-reinigung-berlin | 4480 | 6.49 | 11 |
| gartenbau-munchen | 4130 | 3.58 | 17.6 |
| buero-reinigung-hamburg | 3620 | 8.07 | 27.7 |
| elektriker-hamburg | 2820 | 6.02 | 25.1 |
| elektriker-muenchen | 2270 | 6.27 | 20.6 |
| buero-reinigung-muenchen | 1760 | 12.39 | 5 |
| gartenbau-wiesbaden | 1510 | 2.22 | 26.2 |
| trockenbau-muenchen | 1170 | 4.87 | 32 |
| bodenleger-muenchen | 1060 | 7.21 | 19 |
| fliesenleger-muenchen | 1060 | 3.41 | 37.1 |
| entruempelung-wiesbaden | 820 | 8.42 | 4.3 |
| entruempelung-wuerzburg | 670 | 5.51 | — |
| terrassenueberdachung-berlin | 460 | 5.2 | 44 |
| terrassenueberdachung-hamburg | 460 | 5.8 | 5 |
| entruempelung-potsdam | 430 | 6.02 | 46.4 |
| trockenbau-ulm | 240 | 3.55 | — |
| terrassenueberdachung-muenchen | 230 | 8.5 | 22 |
| bodenleger-ulm | 170 | 3.34 | 44 |
| terrassenueberdachung-ulm | 170 | 3.97 | — |
| wallbox-berlin | 150 | 2.54 | 23 |
| wallbox-hamburg | 130 | 6.18 | 24 |
| bodenleger-rostock | 120 | 1.65 | 43 |
| trockenbau-rostock | 120 | 2.02 | 15 |
| terrassenueberdachung-rostock | 110 | 3.5 | 16 |
| wallbox-muenchen | 110 | 7.08 | 26 |
| fassadensanierung-muenchen | 100 | 6.96 | 29.4 |
| treppenlift-rostock | 100 | 31.63 | 0 |
| treppenlift-ulm | 30 | 73.65 | 8 |
| dachrinnenreinigung-rostock | 20 | 0 | 0 |
| dachrinnenreinigung-ulm | 20 | 0.84 | 0 |
| energieausweis-national | — | — | — |
Wax-to-silver ring kit
TrueToYou.
Active initiatives and their next moves. Pick up where you left to resume work in Claude Code.
Initiatives
4 openUS validation launch
Activeprove demand with real purchases before building full infrastructure. Validation thesis: 10+ committed buyers from ~$1k ad spend . See memory `truetoyou-pricing-gtm`.
0/5 done
Sabri Suby audit fixes (pre-validation)
ActiveAudit at `2026-05-30-sabri-suby-audit.md` found the validation threshold is below the 3:1 LTV:CAC line at current unit economics.
0/11 done
Brand + creative
Tendbrand designed already; ongoing visual polish for ads + site as needed.
capture pad
AI-agent readiness
parkedSame Google IO 2026 / search-agent shift driving the LeadBridge llms.txt rollout. For TrueToYou specifically, agents booking "buy a wax-to-silver ring kit for me" is a plausible di…
0/1 done
The bet
Erbright.
Active initiatives and their next moves. Pick up where you left to resume work in Claude Code.
Initiatives
1 openBuild + launch the Erbright website
ActiveWilli 2026-06-15 — stop validating via cold Kanzlei outreach; just build and launch a working website. Copy already written, brand name decided, domain already owned.
0/5 done
Life outside the businesses
Personal.
Health, fitness, taxes, life admin, nomad logistics. Anything not tied to a business.
Initiatives
3 openPersonal CRM
ActiveWhen a specific problem comes up , having a searchable database of relevant people I know makes it way easier than relying on memory.
0/4 done
Joy as personal health/coaching assistant
ActiveStop rebuilding training context every time Willi wants quick health/training advice. Once Joy has full context , can give expert advice on-the-spot.
0/5 done
Personal / life admin
Tendcaptures personal things that need doing but don't slot under a business.
capture pad
Parking lot
Ideas.
Sparks waiting for Sunday triage. Grouped by business so they're browsable without bleeding into today.
Dream queue
3 freshHermes Agent is the orchestration layer that sits above Claude Code — visual dashboard for activity + memory, unified context across GitHub/Obsidian/Telegram/GCal, with 'dreaming' for proactive self-improvement of skills.
▸ If you implement: Saturday Hermes block: install Hermes locally, point it at ~/Claude/ as the vault, and limit the first session to the dashboard + memory views only — Pantheon and dreaming are separate evaluations that can wait until the base layer feels worth keeping.
Jack Roberts — Hermes NotebookLM bridge (n8n pattern) ↗ai-osproductivity
Jack's video shows n8n wired to a NotebookLM notebook on a schedule, pushing daily personalized insights out — turns NotebookLM from a pull-tool into a proactive briefing engine.
▸ If you implement: Skip n8n (extra runtime), port the pattern: build a launchd job that calls /notebooklm to query a rolling 'intel-context' notebook each morning, writes a 5-bullet brief to ~/Claude/morning-briefings/, and have /morning-brief surface it inline.
Nico | AI Ranking — Pabbly + Claude GBP automation ↗ai-seolead-gen
Nico ships two Pabbly+Claude flows beyond the SEO audits: auto-rewrite blog content into Google Business Profile posts, and auto-respond to positive reviews with negative-review email alerts.
▸ If you implement: Skip Pabbly (subscription-funded path violates runtime-routing rule), port the Claude prompts into two new skills — /lb-gbp-post and /lb-review-triage — gated by leadbridge-pipeline-gate; the GBP-post skill becomes the long-tail content engine for the 13 live sites.
No fresh ideas right now.Come back tomorrow — /dream surfaces new cards each morning.
Operating System
13 ideasactiveJoy OS v1 — lane→initiative→task hierarchy
clean separation between *ideas* (need triage) and *tasks* (already-committed execution); makes weekday execution zero-decision.
activePattern-mining across calls + Claude chats
/granola-digest already mines Granola call transcripts into actionable dream-ideas (the right-side dashboard column). Willi wants the same pattern recognition applied to *Claude chat sessions* — recurring requests, repeated frictions, things he keeps re-asking — and surfaced as automation/improvement suggestions. Same engine, second source: how he talks to me is as load-bearing as how he talks on calls.
activeProductivity OS hardening — post-v1
OS Constitution + audit infra ratified 2026-05-22. Two-week post-build review scheduled per design.
activeSteal-from-gstack — port pressure, not names
gstack (garrytan/gstack) is the most thought-through public Claude Code stack. Six patterns + the artifact-chaining meta-pattern map cleanly to gaps in Willi's OS. Plan approved 2026-05-24 after Codex critique. Full design: `~/Claude/plans/2026-05-24-gstack-port.md`.
activeMap & steal from Jack Roberts' Claude setup
Willi bought into Jack Roberts' paid community (Jack's World Vision). Sibling to the gstack port but a different source — Jack's stack is the thing being mapped here. Goal: reverse-map his entire Claude/AI setup, lay it side-by-side against Willi's OS, and extract concrete improvements + learnings. Public-YouTube-worth-evaluating note from [[feedback_paid_community_creators]] applies; now he's inside the paid community so there's deeper material to pull from.
activeObsidian as substrate over ~/Claude
Convert `~/Claude/` into an Obsidian vault — human UI layer over the existing markdown OS. Doesn't change runtime; unlocks graph view, backlinks, mobile reading, vault portability. Step A of the cadence-engine plan, independent of the runtime migration. Decision context: `~/Claude/shared/research/2026-05-24-always-on-claude-hosting-decision-memo.md`.
activeAlways-on cadence engine on Hetzner VPS
Cadence engine must fire regardless of laptop state — going fully nomadic 2026, deregistering from Germany, no permanent home for a Mac mini. Hetzner ARM VPS + Claude Code subscription OAuth via `claude setup-token` + Tailscale + systemd timers. Full memo: `~/Claude/shared/research/2026-05-24-always-on-claude-hosting-decision-memo.md`. Codex-validated 2026-05-24: runtime stays Claude Code; do NOT migrate to Trigger.dev+OpenRouter as primary.
activeMagier Executive Assistant build queue
Magier is 90% of cognitive load; the EA-build queue is the systematic path to remove Willi from the day-to-day. Sequence settled in hot.md 2026-05-23: JTBD foundation → Granola continuous digestion → feature builds (inbox, LinkedIn, team) → end-state surfaces. The build artifacts live in `~/Claude/` (os-meta substrate even though they serve magier); the *tasks they automate* are tracked in magier's Notion per [[feedback_magier_notion_exclusion]].
pausedEnd-state EA surfaces — DEFERRED parking lot
Three EA end-state visions that depend on the EA build queue producing data first. Captured to remember the destination; not buildable until the feeders exist.
activeM008 morning-brief refactor — code-over-prompt
First concrete application of M008 (code-over-prompt). Extract `/morning-brief` to `morning_brief.py` (launchd 6:55am, top card on dashboard), slim skill to `/override-plan`, then add Override button with Web Speech API → 3 sequential voice questions (displacement / reason / reschedule) → headless `claude -p` → verdict back to dashboard. MVP non-streaming; streaming only if override frequency >5x/week. Full design walkthrough in chat 2026-05-24.
activeiPhone mic — ambient day-recording into Claude context
Highest-level context Claude could ever have — keeps memories continuously up to date without manual capture. From W21 inbox 2026-05-23.
pausedPrint `webflow-pp-cli` via Printing Press
replaces the broken Webflow Designer API + official MCP for magier's workflow. Unlocks (1) Figma → Webflow pixel-perfect conversion without Designer being open, and (2) Slack → Webflow page/section generation against the existing design system so non-technical people can ship pages by chatting. Tool is cross-business reusable but magier is the primary driver.
activeFoundation Sprint skill (`/foundation-sprint`)
Jake Knapp's Foundation Sprint (prequel to the Design Sprint, from "Click") is the upstream clarity/positioning piece the current `/validate-idea` → `/close-bet` pipeline is missing. Build it as a Joy skill, then audit all three businesses (LeadBridge, TrueToYou, Erbright) with it.
magier
1 ideaactiveCustomer & Former-Customer Interviews (ICP Discovery)
Learn who magier's real ICP is and what pain we actually solve, via Mom-Test-style interviews with current + churned customers. Feeds the strategic repositioning bet: move from replaceable "extended workbench" (cattle) to embedded, indispensable partner (pet) — higher CLTV, better case studies.
LeadBridge Local
28 ideasactiveMonthly accounting
recurring monthly bookkeeping for LeadBridge. Willi 2026-06-08: collect last month's invoices + wrap the accounting this week.
activePhone number replacement after EasyBell cancellation
EasyBell is cancelling all four LeadBridge geographic numbers (Ulm, Hamburg, Rostock, Berlin) 2026-05-28 per BNetzA Allgemeinverfügung 25/2006 — a US LLC cannot legally hold Ortsnetzrufnummern. Need a permanent replacement that survives the Firmensitz rule and fits the nomad/no-German-presence reality. Decision memo: `~/Claude/leadbridge-local/research/2026-05-27-german-local-numbers-us-llc/german-local-numbers-us-llc.md`. Path chosen: German mobile numbers (015x/017x) — escape 25/2006, BNetzA Nummernplan has no residency requirement for mobile subscribers.
activeNovel infrastructure / new skills
rolls up `/level-up` candidates as they emerge from Sunday review.
activeAI-agent readiness — llms.txt + Bing WT + agentic-browsing audit
Google IO 2026 shipped browser search agents + personalized dashboards. Click-driven SEO is decaying — measured KPIs shift to citations + search appearances. Cheap, defensible portfolio-wide moves before adoption matters. Source digest: `~/Claude/shared/research/2026-05-25-llms-txt-video/01-digest.md`.
parkedDistrict-level content for agent recommendation
Same Google IO 2026 thesis as the llms.txt initiative — agents know user context (district, postcode, neighborhood) even when the user doesn't type it. "Find me a Bodenleger" from someone living in Prenzlauer Berg gets enriched silently with district context before the search runs. Sites with district-specific content become preferential recommendations even with zero search volume on the district keyword. Source: Willi 2026-05-25, extending the hyper-niche argument from the IO 2026 video digest (`~/Claude/shared/research/2026-05-25-llms-txt-video/01-digest.md`).
parkedLB-improve backlog — PARKED candidates
sparks that are good but not the right thing to build now.
activeBacklog — deferred LB work
see `~/Claude/leadbridge-local/BACKLOG.md` for the canonical list. Top item: EasyBell voicemail-to-email setup. This section mirrors only what `/task` adds via the markdown layer.
tier aTreppenlift
Mature high-CPL niche, aging boomers, well-known buyers
tier aSauna-Einbau / Heimsauna
Longevity wave, B2C + B2B (gyms)
tier aEisbad / Kältekammer Einbau
Emerging, low competition
tier aHeimkino / Wellness-Bad Umbau
High-end renovation
tier aGym & Longevity Equipment
B2B (commercial gyms, hotels, longevity clinics)
tier aBarrierefreier Umbau / Hauslift
KfW-gefördert, aging boomers, big tickets
tier aHörgeräte
Mature market, Geers/KIND dominate but local openings exist
tier aHyperbare Sauerstofftherapie / HBOT Klinik
Emerging longevity angle (frame as wellness/clinic finder)
tier bHaartransplantation
Big market, mature lead gen, Turkey-competition pressure
tier bZahnimplantate
Saturated in big cities, openings in mid-tier
tier bSchönheits-OP
§11 HWG strict, but clinics still advertise heavily
tier bMed-Spa / Botox / Filler
Advertise treatment (not drug name), recurring LTV
tier bIVF / Kinderwunsch
Cross-border (Spain, Czech) also a real angle
tier bLongevity-Klinik / Concierge Medicine
Emerging (Aware, Loft, etc.) — small but high-value
tier bPrivatklinik Alkoholentzug / Sucht
Oberberg, My Way Betty Ford, Eternity Recovery as buyers; €15–30k stays
tier bAdipositas-Behandlung (GLP-1 angle)
Frame as consultation/clinic — NEVER name Rx drug. ZAVA, DoktorABC, Adipositaszentren as buyers
tier bACP / Eigenbluttherapie / Regenerative Ortho
PRP/ACP fully legal, private Orthopäden are buyers, cash-pay €500–5000
tier bAugenlasern / LASIK
Cash-pay, established lead gen market
tier bTRT / HRT Beratung (consultation framing)
Cannot advertise testosterone directly — route to endokrinologen/urologen for consult
tier bHormonberatung Frau / Menopause
Trending — Mary Claire Haver wave reaching DE
tier bKetamin-Therapie Depression
Off-label, small market, but emerging
TrueToYou
4 ideasactiveUS validation launch
prove demand with real purchases before building full infrastructure. Validation thesis: 10+ committed buyers from ~$1k ad spend (US-first, pricing $279 USD). See memory `truetoyou-pricing-gtm`.
activeSabri Suby audit fixes (pre-validation)
Audit at `2026-05-30-sabri-suby-audit.md` found the validation threshold (10 buyers / $1k = $100 CAC) is below the 3:1 LTV:CAC line at current unit economics. Five offer/copy fixes must ship before paid traffic — otherwise the $1k is burned with no clean signal.
tend modeBrand + creative
brand designed already; ongoing visual polish for ads + site as needed.
parkedAI-agent readiness
Same Google IO 2026 / search-agent shift driving the LeadBridge llms.txt rollout. For TrueToYou specifically, agents booking "buy a wax-to-silver ring kit for me" is a plausible discovery channel post-launch. Source digest: `~/Claude/shared/research/2026-05-25-llms-txt-video/01-digest.md`.
Erbright
2 ideasactiveBuild + launch the Erbright website
Willi 2026-06-15 — stop validating via cold Kanzlei outreach; just build and launch a working website. Copy already written, brand name decided, domain already owned. Deliberately minimal branding: no full brand system, no social/other brand assets. The deliverable is a functioning, launched site — nothing more.
parkedGriefGuide — All-in-One Funeral / End-of-Life Platform
Families facing a death spend 40+ hours juggling funeral arrangements, paperwork, and notifications while grieving. GriefGuide consolidates every end-of-life task into a single, calming dashboard that guides families through the entire process. Users can compare local funeral…
Personal
3 ideasactivePersonal CRM
When a specific problem comes up (Erbright legal contacts, biz dev, technical), having a searchable database of relevant people I know makes it way easier than relying on memory. Concrete trigger: needing the right lawyer for Erbright right now. From W21 inbox 2026-05-23.
activeJoy as personal health/coaching assistant
Stop rebuilding training context every time Willi wants quick health/training advice. Once Joy has full context (food, weight, training routine, mobility, handstand goals), can give expert advice on-the-spot. Previously was using ChatGPT and constantly re-explaining training context. From W21 inbox 2026-05-24.
tend modePersonal / life admin
captures personal things that need doing but don't slot under a business.