Developers work significant untracked hours on subscription accounts, making it impossible to construct an accurate cost-overrun argument before commercial renegotiation calls. · Seen in: 2026-07-17-1100-check-in-open-topics-upSa.md, 2026-07-17-1059-check-in-open-topics-8TWb.md

An n8n workflow that runs nightly, queries the time-tracking tool for all subscription accounts, compares logged hours against the sold pipeline capacity, and posts a Slack digest to the account PM listing any account where logged hours exceed 90% of capacity or where the ratio of unlogged-to-logged hours exceeds a threshold — flagging accounts that need a manual hour reconstruction before the next client call.

Build a /capacity-overage-sweep skill that accepts an account name and a billing period. n8n pulls logged hours from the time-tracking API, compares against the sold capacity stored in Notion per account, and for any account in overage posts a structured Slack message to the PM with: hours sold, hours logged, estimated unlogged hours based on developer self-report, and a one-click link to the account's Notion page. The unlogged-hours estimate uses a simple heuristic: if a developer is listed as active on the account in the project management tool but has fewer than 2 logged hours on a given day, flag the day as potentially untracked. The PM reviews and corrects before the client call. No automatic billing action is taken.

Anna had to reconstruct Blume's cost overrun from memory and developer self-reports the night before a commercial renegotiation call; if the Tuesday call goes badly, the absence of a defensible hour log weakens the rebuild-or-exit argument and risks a worse commercial outcome.

QA on staging environments misses cookie-consent-dependent JS functionality because the staging domain is never added to the Cookiebot account, causing go-live failures that only surface when the client is watching. · Seen in: 2026-07-22-1205-otp-monthly-8rhv.md

An n8n workflow triggered when a new Webflow project is created in the project management tool that automatically adds the staging domain to the client's Cookiebot account via the Cookiebot API, and posts a confirmation to the project's Slack channel so the PM can verify before QA begins.

Build a /cookiebot-staging-register skill that accepts a Cookiebot account ID and a staging domain URL. n8n listens for new project creation events in Many Requests or Notion, extracts the staging URL from the project brief, calls the Cookiebot API to add the domain to the account's allowed list, and posts a confirmation message to the project Slack channel. The Cookiebot account ID is stored per client in Notion. If the client does not yet have a Cookiebot account, the skill posts a flag to the PM instead of failing silently. Pilot on the next OTP project after Vivanta Homes.

Alessa named this exact failure on Vivanta Homes — cookie-consent-restricted JS only caught at go-live — and it is a repeatable risk on every future project that uses Cookiebot; the fix takes one API call and costs nothing to automate.

Design fixes applied in Chrome or Firefox are not re-verified in Safari before client delivery, causing visual regressions on Safari that clients catch at go-live and attribute to poor process. · Seen in: 2026-07-22-1205-otp-monthly-8rhv.md

A Claude agent that runs when a developer marks a ticket as ready for QA, reads the ticket type, and if the ticket involves any visual change posts a structured Safari QA checklist to the ticket in Many Requests — listing the specific elements changed and requiring a Safari screenshot attachment before the ticket can move to client delivery.

Build a /safari-qa-gate skill that accepts a ticket ID. When a ticket status changes to ready-for-QA in Many Requests, n8n triggers the skill. Claude reads the ticket description and change log, identifies the visual elements modified, and posts a checklist comment to the ticket listing each element with a checkbox and a required Safari screenshot field. The ticket cannot be marked complete until all checkboxes are ticked and at least one Safari screenshot is attached. The checklist is generated dynamically from the ticket content, not a static template. Pilot on all Webflow OTP tickets for the next two weeks.

Alessa named Safari QA as the specific failure mode on Vivanta Homes, and the client escalated it into a broader process criticism that reached Willi; a second Safari regression on any account in the current churn-risk cohort would accelerate the departure of clients already identified as at-risk.

Sales calls following webinar sign-ups require Gabriel to reconstruct ICP, pain points, and creative direction from scratch in each call, with no structured brief passed from the webinar lead to the creative team before the follow-up meeting. · Seen in: 2026-07-17-0801-magier-webinar-demo-between-maximilian-fleitmann-and-emmanue-KWBw.md, 2026-07-20-0759-kristian-x-magier-6U8f.md, 2026-07-22-1200-emma-lagast-x-magier-sBcI.md

A Claude agent that runs within 30 minutes of a sales call transcript being created in Granola, extracts ICP, stated pain points, competitor references, creative direction signals, and agreed next steps, and posts a structured one-page brief to the client's Notion page and the relevant Slack channel — ready for the creative team before the follow-up meeting.

Build a /sales-call-brief skill that accepts a Granola transcript URL or basename. Claude reads the transcript, extracts six fields: company name and ICP, stated pain points, competitor references, creative or brand direction signals, agreed deliverables, and follow-up meeting date. The output is a structured Notion page created in the client folder with these six fields populated, plus a Slack message to the sales channel tagging Gabriel and the assigned PM. The skill runs automatically via n8n when a new transcript appears in the Sales Calls folder in Granola. No human action required; the PM reviews and corrects before the follow-up call. The three webinar-to-demo calls this week each required Gabriel to re-extract the same information manually during the call.

Gabriel is running multiple free-work-sample cycles simultaneously — SIIT, Serpier, Pitchy — each with a tight turnaround before a follow-up meeting; if a brief is missing or incomplete when the creative team starts, the 48-hour turnaround promise fails and the free sample loses its conversion power.

OTP clients commit to feedback deadlines verbally at kickoff but miss them without consequence, causing design resources to sit idle or be reallocated, then creating a rush when the client re-engages and expects to resume their original queue position. · Seen in: 2026-07-22-1205-otp-monthly-8rhv.md

An n8n workflow that monitors open OTP tickets in Many Requests for feedback-pending status, sends an automated reminder to the client contact 24 hours before the agreed feedback deadline, and if the deadline passes without a response posts a queue-position warning to the client and a Slack alert to the PM — making the consequence of missing the deadline explicit and visible before it becomes a resourcing crisis.

Build a /feedback-deadline-monitor skill that runs as an n8n cron every morning. For each open OTP ticket in Many Requests with a pending-response status and a go-live date set, Claude calculates whether the feedback deadline is within 24 hours or already passed. If within 24 hours, n8n sends a templated reminder email to the client contact via the transactional email provider. If the deadline has passed, n8n posts a Slack alert to the PM with the ticket name, days overdue, and a pre-drafted client message stating that the project will be re-queued if feedback is not received within 48 hours. The PM approves the message before it is sent. The go-live date and feedback deadline are set at project kickoff and stored in the ticket. Cameron named Legion Security, Happy Client, and Ariba as current examples of this pattern.

Cameron is managing three simultaneously stalled OTP accounts while also absorbing 2-4 hours of internal work daily; each stalled account that re-engages without warning consumes design capacity that was reallocated, and the resulting scramble is the direct cause of the process uncertainty he named as his primary stress driver.