What needs to move. Not what is already done.

Three shipments in sequence (2026-05-20): (1) commit 875174a: 'Show LINE ID' checkbox + input in TH Sarabun (yesterday's work). (2) commit 00c11ed: first font-load fix attempt for save-as-PNG (Sarabun preload only). (3) commit f403926: real root cause for Sotheby's save-PNG bug was toPng's backgroundColor option overriding the paradigm's own dark teal background, leaving CREAM text on CREAM canvas. Removed the option, expanded font preload to Bricolage + Manrope. ALSO shipped the generic free-text 'message / description' box Mameaw asked for in her 2026-05-19 email plus Wil flagged at 03:51 BKK: TH Sarabun, multi-line via whiteSpace pre-line, renders on all 4 paradigms with correct color flip (CREAM on dark Sotheby's + Compass left, INK on Knight Frank + Luxury + Compass right). Live: https://dlvxrentals.com/property-picker. 1.5h logged.

Wil brief 2026-05-20 10:08 BKK: contrast 2mo of paid outbound (SDR EUR 3500/mo, 2000 LI msgs, Instantly list, hours pulling, 5 offers pushed, result 3 calls + 1 partner convo) vs 2mo of inbound from doing-the-work (Thai retailer app shipped -> 3 inbound, Belgian client happy -> bigger project, ex-colleague referral with 10% finder fee -> deal closed). Result: EUR 80K pipeline. Frame: integrity-built inbound beats burned-cash outbound. Raw register, no AI tells, ready-to-go viral. Draft below (one body, 3 closes). --- BODY --- Two months ago I hired an SDR. EUR 3,500 a month. Two months in. - 2,000 LinkedIn messages sent - Instantly emails fired through a scrubbed list (hours pulling, verifying, segmenting) - 5 offers pushed - Result: 3 calls. 1 partner conversation. Two months of just doing the work and being in the room. - Shipped an app for a big retailer in Thailand. 3 people reached out asking me to help with theirs. - Finished a project for a Belgian client. They're so happy they handed me a bigger one. - Someone I used to work with referred a deal. He gets 10%. Closed. Pipeline today: EUR 80K. One side of that came from cash I lit on fire. The other side came from people who saw me ship and trusted me. --- CLOSE A (lesson hammer) --- The thing you're least excited to do (post the case study, ship the work in public, thank the referrer like a human) is the thing that compounds. The thing you can pay a stranger to do at scale? Doesn't. --- CLOSE B (provocation) --- Stop hiring SDRs. Start being someone people want to introduce to other people. Took me EUR 20K in lead gen and 87 freelancers to figure that out. --- CLOSE C (anti-tactic) --- Nobody is going to refer the agency that DMs them 2,000 times. They'll refer the one whose work they remember. Build the second one. The math will sort itself.

Bella replied to Jules MOUCHET (RSDoutdoor) 2026-05-21 from hello@piexels.co with the WP admin path (Products > All Products > SP-56 > Product data, swap Add new for Add existing). Kristof was CC'd on that send. Loop on Jules side is closed.

Sara sent two batches: May 19 20:58 (13 NL/FR/ENG fixes plus questions) and May 20 17:25 (12 DE text swaps plus 2 full DE translations in an .odt attachment). Bella consolidated everything to ~/Desktop/epic-hottlet-feedback-consolidated.md (Section A bugs, Section B DE text swaps, Section C DE translations, Section D SEO retainer plan for closing email). Workflow: Wil opens Claude session against the doc and executes section by section. Old Phil-triaged plus Sara-reply-DRAFT framing is superseded.

Patrick flagged 3 nits, 2 of 3 fixed by Phil 2026-05-12. Wil corrected 2026-05-21: sticky CTA was already added. Do not keep this as a build task until the live Framer page is checked. Invoice 4KSVLJL2-0001 (EUR 8K) remains an AR follow-up, but the task is now verification plus payment follow-up, not 'Wil still needs to add CTA'.

Site is live but invest page still missing env vars. Previous board incorrectly named Louis. Wil corrected 2026-05-21: there is no Louis on Momoamo. Treat owner as unknown until the client truth file and repo env docs are checked.

Google Workspace piexels.eu payment failed 2026-05-20 18:51 (notices hit wilcot@piexels.eu plus hello@piexels.co). Instantly.ai 2nd unsuccessful-payment notice 2026-05-19 11:39 on hello@piexels.co. Wise occupation-confirmation KYC form 2026-05-19 22:50 on hello@piexels.co (account hello@studiodlvx.be). All 3 block outreach or payouts if ignored.

Calendar invite Johan Wouters (johan@nemesis.studio) 2026-05-20 16:26. Google Meet, 15 min. Wil's angle: position Piexels plus Nemesis as a duo offer (Nemesis brand, Piexels dev). Already informal via margin-add pattern (kan nemesis marge er bij doen). Prep brief in Telegram msg 4599.

Wil's thesis 2026-05-21 10:00 BKK: value-based not chronology-based. v1 at ~/Desktop/wil-resume-v1.md (7 bullets: agency at 25, 5 apartments 3 sold 80%, 700K agency revenue, entrepreneurial mindset, Belgian government website, masters 80% remote during 2 jobs, Russian in 1 year). Application flow doc at ~/Desktop/wil-resume-application-flow.md covers sourcing channels, personalization, follow-up cadence, 10 more angles to rotate.

Phil audit (live PageSpeed): mobile Perf 58, LCP 9.9s, FCP 7.2s. Vimeo iframe on hero is the dominant cost. Lazy-mount via IntersectionObserver or poster+play-button swap. Biggest single SEO lever on epic.be.

Cookie + privacy NL/FR 404 (CMS seed missing), social buttons missing in footer (RLS read), fish icon swap to 'quote to come' placeholder, locale-aware webshop URL (conditional), recipe subtitles i18n strings NL+FR. Phil holds the diff, deploy pending approval signal.

(a) Who-we-serve: all 3 subsections editable not just Retail. (b) Retail IFS brokerage copy field opened (currently AI placeholder). (c) 2027 certification: confirm CMS path or add field. Drafted as separate Hilman ticket if Phil can't do directly.

Sara wants Hottlet logo more visible in header + footer. Asset (PNG/SVG transparent) not yet supplied.

Algemene voorwaarden currently EN-only on NL+FR routes. Sara to supply translated text.

Recipe missing in NL+FR. Either Hottlet translates in-house (free, their pace) or Piexels via paid translator (~EUR 150 per language).

Sara asks where Hottlet sources the Meta Pixel ID. Two paths: Sara creates Pixel in Meta Business Suite (free, ~10 min) or Piexels sets it up one-off (billable hour).

Currently EN-only by schema. Sara expects NL+FR. Adding the columns + admin form + i18n picker is ~4-6h Hilman work at EUR 65/h.

Phil pulled live: 226 clicks, 4,874 impressions, 4.64% CTR over 26 days (mostly branded). epic.be marketing 113 clicks 7.36% CTR. Indexation clean. Top query 'hottlet frozen foods' 52 clicks. Mobile Lighthouse 58 (Vimeo hero, see item above). Report saved to studio/clients/epic/reports/2026-05-20-seo-retro.md. Will attach to Sara reply. Bi-weekly cadence kicks in if Hottlet signs the SEO retainer (proposal still unsigned).

Sara confirmed German version was sent for review, she'll send feedback ASAP.

Sara screenshot attached. Cookie in NL+FR and privacy in NL+FR all 404. Other locale URLs work. Routing or static-export bug.

Sara added the social links in CMS backend but the footer does not render them. Likely the footer component is not bound to that CMS field, or field name mismatch.

Sara's colleague says fish icon must NOT be used. Currently rendering as placeholder. Needs swap to neutral placeholder or removed entirely.

Brand visibility. Logo should appear in header at minimum, ideally also footer and OG image.

Layout regression or intentional but reads as empty. Sara flagged.

Sara asks if mixed sizes are by design or a bug. Also asks if everything is in the IvyEpic font.

Currently the webshop link is the same regardless of site locale. Sara expects locale-aware destination.

Sara says the description is wrong, they cannot edit it, the field is filled but reads as empty. She suspects AI placeholder copy. Backend only allows editing the retail section.

Sara wants to edit all sections, currently only retail is editable. CMS schema gap.

Sara expects FR and NL versions. Currently both show EN content.

Recipe template subtitles still English on NL and FR locales.

Sara asks who handles the translation. Decision needed.

Sara asks if product titles will get translation support or stay EN.

Hottlet has a new certification with 2027 validity. Sara asks where she edits this herself.

Sara asks where to source the Meta Pixel ID for tracking setup.

Sara asks when the first bi-weekly SEO report arrives. Brain check: there is NO signed Hottlet SEO retainer, only an unsigned proposal pitch. The bi-weekly cadence was conditional on signing. The right answer to Sara is the proposal, not a delivery.

Sara confirmed German version was sent for review, she'll send feedback ASAP.

Stripe live: status=open. BEHVA insurance site for GEXIA SRL. Tied to sticky CTA fix above. Renaud aware of due date but no nudge sent since May 7.

Stripe live: status=open. Real client, real invoice (Wil confirmed 2026-05-07). Payment plan in motion via Wise installments. Stripe shows full balance until Wil marks partial paid.

Previous board framed this as uncollectible and suggested write-off or direct pursuit. Wil flagged that framing as too aggressive and unclear on 2026-05-21. Do not label it as a final loss until Stripe current status, touchpoints, and Enky relationship context are checked live.

Stripe live drafts were listed as Debussche 206.25, Hottlet 675.00, Reynaert 150.00, Weblounge 262.50. Wil corrected 2026-05-21: Debussche and Hottlet drafts should be deleted if they are still drafts. Reynaert and Weblounge need separate verification.

Wil 2026-05-21: piexels.dev/co becomes piexels.co. Sub-branches go to th.wilcot.co and be.wilcot.co. Run order: Wil walks /co manually for correctness and mobile, captures issues via the deployment feedback tool, then spawns Phil session to ship the migration. Then check be and th. Push SEO plus GEO hard. Not 100 percent optimize until real traffic.

Schema, sitemap, robots, canonical tags. Audit drift across the be/, th/, co/ locale routes. Existing audit at /audit endpoint can run live.

Wil 2026-05-21: 'we need more tools for SEO, for all websites in our control, understand better and grow more'. Stack candidates: Ahrefs (best backlinks, EUR 99 to 449/mo), Semrush (best keyword and competitor, EUR 119 to 449/mo), SE Ranking (budget, EUR 65 to 189/mo), Surfer (content optimization, EUR 89 to 199/mo), DataForSEO API direct (pay per call, cheapest at scale). Already running: Search Console, GA4, Plausible on some sites.

Clutch is the agency-tier surface for piexels.dev (peer-reviewed B2B). Per-client Senja accounts isolated by domain (HT already separated). 5 HT review requests sent 2026-05-13, await responses.

5 named studies (enky, momoamo, jobfin, qlyr, nemesis). Want screenshots, scope, outcome, before/after metrics. Some already drafted in content/case-studies/.

unrealgymthailand.com is the public site. Wil walks home, classes, coaches, membership, contact, reset-club, gear. Flags issues per page via the deployment feedback tool. Phil and Hilman pick up the list.

Strategy doc at studio/clients/unrealgym/app-strategy.md (React Native, Expo, Supabase, Stripe, OpenAI). Tonno is the MD-side contact for asset handover. Booking system is the priority workstream inside the MVP, see next item.

Class booking (group classes, time slots, capacity caps) plus personal training (coach availability, slot booking). Integrates Supabase data with Stripe per-session billing for non-members.

Arcade-prep in progress. Queued: weblounge, boi, unreal, momoamo, behva+newlands, epic, horsetonic, enky, kymono. THREAD.md is the one timeline per client. Comms folder deprecated.

When a new client lands: studio/clients/[slug]/ folder created with identity.md, THREAD.md, comms/ (read-only), docs/, tasks/. Hilman briefs forced to template. Time logged to operations/timesheets/YYYY-MM.md.

HT Senja was almost signed up under hello@piexels.co. Rule now in .claude/rules/client-tool-isolation.md. Need pre-flight checklist before any tool signup on a client deliverable.

Rule exists in client-work.md but compliance drifts. Want a Friday Otto check that flags client work without matching timesheet entries.

2 months. 3500/mo. Hired an SDR. She sent 2000 LinkedIn messages. We pulled lists. Wrote sequences in Instantly. Pushed 5 offers. Result. 3 calls. 1 partner. In the same window: Built an app for a big Thai retailer. 3 people DM'd asking for help. Finished a project for a Belgian client. They liked it. Now they want a bigger one. Old colleague sent me a lead. Closed it. He gets 10%. Pipeline from outbound: 1 partner conversation. Pipeline from doing the work: 80k. I am not anti-outbound. I am anti-pretending it scales when you are solo and not Stripe. Trust pulls more than scripts.

Tracked two columns this year. Column A. Outbound. - 1 SDR at 3500/mo for 2 months - 2000 LinkedIn messages sent - List-pulling and copywriting hours stacked - 5 offers out - Result. 3 calls. 1 partner. Column B. Doing the work. - Built an app for a big Thai retailer. 3 inbound conversations the same week. - Shipped a project for a Belgian client. They came back for a bigger one. - A colleague I worked with years ago sent me a lead. Closed it. 10% to him. - Result. 80k pipeline. Same calendar. Two strategies. One closed itself. Outbound is a casino when you are solo. The work showing up for you is the work showing up for you.

My SDR clicked send on her 2000th LinkedIn message last week. Same week I shipped an app for a big retailer in Thailand. The next morning 3 people I had never met asked if I had bandwidth for theirs. A Belgian client I just delivered emailed: we have something bigger now. A guy I used to work with sent me a project. Closed it. He is getting 10%. The SDR run cost 7k. Two months. 5 offers pushed. 3 calls. 1 partner. The just-do-good-work run? 80k pipeline. I am not saying outbound is dead. I am saying for a solo operator, the leverage is the wrong way around. You cannot out-script someone who already trusts you.

Samah replied to Wil's comment in her Meta-AI-logo thread quoting him: 'Consistency is execution, not strategy.' Public thread. Wil's call: reply in thread, DM, or skip.

Old drafts deleted Apr 23. Need new lineup. Anchors: Dubai shutdown, EUR 20K lead-gen lesson, ADHD ops, Hilman async stack, piexels.dev launch arc.

Per .claude/rules/linkedin-comments.md: max 25 words, max 2 sentences, casual register. Every comment drafts to Telegram for approve/edit/skip. No auto-posts.

Likes are autonomous (25/day). Birthday/anniversary DMs autonomous. Comments draft-only. Connection requests on ICP per Mango research.

horsetonic daily 7AM, linkedin weekdays 5PM, upwork daily 4PM, otto Tue/Fri/Sun 6AM, bella-am daily 7AM, bella-pm daily 6PM, finn Mon 9AM. Failure seeds at operations/seeds/.

Bella bulk-cleared 12 noise entries today (rule files, skill docs, content drafts, briefing templates, QA reports quoting 'I will' or I'll as example text, not actual outbound promises). Same pattern was cleared en masse on May 13 and May 16. Hook keeps re-capturing example text from .claude/rules/ and projects/piexels-co/sections/. Audit trail in operations/state/reality-clock.yaml header.

Snapshot file lags by days. Per .claude/rules/client-project-status-from-live.md: state.md is a hypothesis, not truth. Live APIs win.

GA4 secret item resurfaced as 'still open' in morning sweep but was already done by an agent yesterday after Wil granted Google API access. Wil flagged. Rule now codified at .claude/rules/verify-action-items-live-before-resurfacing.md. Hook precondition proposed.

Two new rules: refero-mcp-mandatory.md (pull references first) plus impeccable-design-skill-mandatory.md (catch AI-slop). Every web-build agent needs both in boot prompt.

Anthropic Agent Teams native (flag enabled, mailbox messaging, shared task list). Ruflo orthogonal (SONA learning, 100+ pre-built agents, 30-50% token savings).

MRR baseline EUR 1,000-1,500 from Enky retainer plus hosting. To hit 10K: 3 clients at EUR 3K/mo retainer. Top warm: La Superboite (next check Jun 24), FCKLCK/James, Mucho/Marina.

Per .claude/rules/proposal-rigor.md Rule 3b: when Wil says 'those big agencies', map the full tier. Ostmodern, Crema, ThoughtWorks, Globant, EPAM, R/GA, AKQA, Work&Co, DEPT.

Per .claude/rules/proposal-rigor.md Rule 3c: paint upper-bound version of every public surface. Press page, team page (real composite), case-study videography, certifications-in-progress framing.

BOI route shelved 2026-05-14. Active path: 49/51 with Mameaw. v7 confirmed acceptable for both Thai and English signing per Konrarat 2026-05-19 22:21. Bella shipped v8 final PDF (already signed by Wilcot as First Shareholder) to konrarat.deesawat@gmail.com 2026-05-20 09:58 BKK. Awaiting her countersign.

Sophie replied 2026-05-20 17:05 CEST: declining Phase 1 to run with their own dev for 4 weeks, BUT leaves the door open for Phase 2 follow-up ('De kans blijft reëel dat we ons vast fietsen... Blijven jullie openstaan voor een eventueel vervolg?'). Wil sent his own warm reply 2026-05-21 keeping door open with a 'let us know in time so we can hold capacity' scheduling note.

WhatsApp soft nudge sent 2026-05-11. Brain says ball in her court. Unipile WA index does not surface her thread, so live verification is partial.

Call Apr 14. Email Apr 16. Wil: no follow-up for 2 months. Next check Jun 24.

Wil transferred EUR 600 to DLVX Invest zichtrekening 2026-05-18 21:31. Board previously framed Jeremy as pushing back. Wil flagged the summary as confusing on 2026-05-21. Treat as unread source-of-truth required. Jeremy confirmed receipt and appears to point to a formal process for the EUR 300/mo reduction, but this must be re-read before action.