Co-hosted by Coach Lucy & Dr. Rob. Rob: this one is about giving you the keys.
Our virtual team is now organized like a company: two human co-founders, a named AI C-suite, department leads, and the functional teams that do the work. And Dr. Rob's own Claude or ChatGPT can now send modules straight into our automated pipeline. Rob approves the preview; the system does the rest. Lucy is no longer the middle step.
Co-hosted by Coach Lucy & Dr. Rob. Rob: this one is about giving you the keys.
We did three things. We reorganized the whole AI team into a readable org chart with faces and function badges. We added Dr. Rob's Production Assistant, a team member that takes his modules from his own Claude session to a live page. And we built the direct line: a Production Requests system that Rob's Claude, ChatGPT, or a simple web page can submit to, with an automated pipeline, a private preview, and Rob's own approval before anything goes live.
Until today, every module Dr. Rob built followed the same path: he finished it in his Claude session, dropped the files into the Dr Rob Input folder, and waited. Lucy then had to notice the drop, open a new project, run the Activator, brand it, gamify it, wire the Clipboard, deploy it, and report back. The work was good, but it all ran through one person's calendar.
Rob ships fast and ships big. The limit on how much of his work reached members wasn't Rob and it wasn't the tools. It was the hand-off.
| Area | Before | Now |
|---|---|---|
| How Rob hands off work | Drop files in a shared folder; Lucy starts a new project | "Send this to my Production Assistant" from his Claude or ChatGPT, or drag-and-drop on the Production Requests page NEW |
| Who runs the pipeline | Lucy, manually, when she has time | The production queue, every hour (or on demand), using our own skills NEW |
| Review | Lucy reviews, then tells Rob | Rob gets a private preview link and approves or requests changes himself NEW |
| Going live | Lucy runs a deploy script and pushes | Approved work is published automatically and logged to all three ledgers |
| After go-live | Fixes come back to Lucy by email or chat | Rob does the post-deploy check and sends fixes as change requests |
| Status visibility | Ask Lucy | Every request has a timeline and activity log; Rob can ask his Claude "what's the status?" |
| The AI team | A list of bots and workflows, names mixed with processes | A C-suite org chart: executives, department leads, functional teams, diverse avatars with function badges NEW |
The rule behind the chart: a name means someone is in charge. Every function has one named AI executive. Department leads are the personas people talk to or call by name. The functional team is the set of workflows and SkillBots that do the work, listed by what they do. Every face wears its function icon as a badge, so a glance tells you both who and what.
Executive names are proposals in the style of Carl, Maggie, Sally and Vicky (Frank Finance is confirmed). The full, clickable chart lives at /super-admin/ai-team/.
One system, three front doors, and the same automated pipeline behind all of them.
Rob adds the FFH Production Team connector once. After that, "send this module to my Production Assistant" submits it, "what's the status?" checks it, and "looks good, approve it" sends it live.
The same connector works in ChatGPT, or Rob can use a private FFH Production Assistant GPT. Files he uploads in that chat are passed straight through.
/production-requests/: pick a team member, drag in files (up to 500 MB each), tick the pipeline steps, send. Track, approve and comment on the same page.
Nothing goes live without the requester's approval. Previews are always gated and hidden from search. Deploys refuse to run if a gate goes missing. Each person has their own revocable key, stored only as a fingerprint. The database only lets requesters approve, request changes, or close; everything else is written by the production team. The master Supabase key stays on Lucy's Mac and in Vercel, never in the shared folder or the code.
A physician co-founder can now take a module from idea to live page without an engineering or content team in the loop. Every new contributor, whether an author, a Chamber Director or an intern, can get the same direct line.
The request portal, the pipeline, the connector and the org chart are ours, on the same lean stack (Vercel, Supabase, Claude API). No project-management SaaS, no agency, no chatbot vendor.
The same system accepts work from Claude and from ChatGPT. We're not tied to one AI vendor, and the standards that make a page "FFH" live in our own skills.
The Production Requests pattern (submit → automated build → preview → approve → live) is the same thing partners ask for: the Omni Media Maker, Chamber of Health Directors, and white-label partners like SCFA.
Coach host: co-hosted. Dr. Rob fronts the "physician-to-platform in one step" story; Coach Lucy fronts the lean-team and investor story.
Do the four setup steps, then have Rob send one small module through all three doors (web, Claude, ChatGPT) as the pilot.
Once the pilot passes, retire the "drop it in the folder" habit. The Production Requests system becomes the standard way work enters production.
Approve or swap the executive names, then write each executive's job description so the team behaves consistently.