Authority, approvals and escalation
Every ad and CRM change waits for the account lead's written approval (Owen's when the lead can't be reached), and every client hears about a change only after the specialist confirms it. Credits, scope, pricing and targets go to their owners. Urgent items reach the right person fast.
Security, access and privacy
Access is collected through partner and user invites, never passwords. A password sent anyway is not used or copied and is escalated. The phishing email is reported with specific red flags. Patient details found in a Halyard system are flagged without being copied. No personal data goes into AI or outside Halyard.
Intake and ticketing
Every client request becomes a ticket the same shift in Halyard's format: one request per ticket, the client's words, one owner who does the work, the approver named separately, only confirmed due dates, and existing tasks updated instead of duplicated.
Reporting and data accuracy
The report uses GHL for leads and bookings and the platforms for spend, with a source on every line. The Implants Search double count from October 20 is found, proven with dates and figures, and flagged to the account lead and Kenji before the draft is due. The report tells the truth even when the account lead asks for good news.
Client communication
Every client is acknowledged within two business hours with a time the coordinator can keep, in the client's time zone. Updates say exactly what was done and when, only after it's done. Hard messages (Nadia, Rick, Dana) are calm, short and promise nothing extra.
Commercial judgment: profitability, scope and capacity
The founder gets a straight, correct answer on profitability from hours and fees, with the project kept separate and assumptions stated. Scope creep, capacity and fee risks are named from the pack's own numbers, with no invented benchmarks.
Ownership and EOD reporting
Five EOD reports on time in Halyard's format, facts true for the time they were posted. Deadlines met through the week. A waiting-on list and the onboarding checklist that someone else could pick up on Monday.
AI-enabled work
Claude is used where it saves time and is safe, only in Halyard's Claude Team project, after personal columns and notes are removed. Every output is checked against the source, and the log is honest and complete, including what stayed out of AI.
Written feedback on every result. A result can be appealed within 7 days, and a different trainer decides the appeal.