Plays
0across 4 outcomes + entry + setup
Pick a play, follow the playbook, use the AI skills that power each step. Plays are 12-month commitments, adjusted quarterly, reported monthly, executed weekly via sprints.
This portal is the front door to the BB AI Workforce. The plays in §02 are the work; the skills in §03 are the AI that powers it. Find yourself in one of the three scenarios below — that's where to start.
The portal reads from a markdown library. Every play in §02 is a real .md file in /playbooks/ — edit the file, refresh the page, the change is live. Same for skills in §03 (each is a folder under ~/.claude/skills/ with a SKILL.md).
You're an EP, CM, or TM picking up this week's work for a client.
/BB Clients/[Name]/03-Plays.bb-voice-check against the client voice profile before you ship.Stack · Claude Code · Granola · HubSpot · client folder
You're an AM at week 4 of onboarding, or kicking off a new quarter.
02-Strategy/outcome-Q[n].md.Stack · HubSpot · Granola · Strategy folder · Claude Project
You're an AM or Strategist running the day-90 / day-180 / day-270 review.
04-Reports. Map to the play's "definition of done."Stack · HubSpot reporting · 04-Reports · 02-Strategy
Two entry products. One setup play. Then 36 plays organized into four outcomes — Attract, Convert, Close, Customer. Each quarter the AM and client commit to one outcome and run tier-allocated plays against it.
Plays are how clients buy and how the team delivers. Click any tile to filter §02 to that play.
Each playbook step calls a specific AI skill. Live skills are ready today. Planned skills are coming in waves 2 and 3.
Every BB skill enforces these. If output violates one, the skill needs fixing.
Thoughtful operator talking to a peer. Not a vendor. Not a guru.
Use a regular dash with spaces ( - ) or a comma.
Default to short. Vary length only for rhythm.
"In today's fast-paced world." "I hope this email finds you well." Banned.
Use [TK: source needed]. Never fabricate.
HubSpot, SearchAtlas, Replit, Claude. Not "a CRM."
StoryBrand, EOS, Working Genius, PEACE Sound Bites.
Every deliverable closes with one bolded Next: line.
"I don't know" beats a confident hallucination.
Where files live, how to use this, what to do when something is wrong.
Every client gets a standard folder structure. The playbooks all reference these paths.
/BB Clients/[Name]/ ├── 00 — Brand design system, voice, logo (Claude Project) ├── 01 — Calls Granola exports + Doc summaries ├── 02 — Strategy 12-month plan, QBRs ├── 03 — Plays/[PlayName]/ drafts/, final/, live-assets.md ├── 04 — Reports monthly performance, QBRs ├── 05 — Sprints recaps, weekly outputs └── 99 — Archive
Plus: Granola folder per client (named the same), Claude.ai Project per client (brand assets + voice loaded as Project Knowledge).
The playbook content lives in markdown files at /docs/bb-operations/playbooks/. One .md file per play. Edit the file, save, refresh this portal — changes appear immediately. The portal is just a visual front end for the markdown library.
If you add a new play, also add an entry to playbooks/index.json so the portal picks it up.
A play is a 12-month commitment we make to a client. Foundation tier runs 1 play at a time, Growth runs 2, Scale runs 3+.
A sprint is the weekly execution unit (prep → meeting → execute → async). Each play moves forward a few sprints per quarter. The playbook tells you what each quarter's sprints should ship.
Open the skill card in §03, copy the prompt, paste into your Claude tool of choice — Claude.ai (with a Project for repeated use), Claude Code (auto-loads if installed), or as a one-off message. Always run bb-voice-check on output before sending to a client.
Tell Jay or Brandon in #ai-workforce. Skills with willing human owners (you'd sit for the 30-min interview + review the first 5 outputs) ship faster.
That's the point of quarterly review. Each playbook has "Adjustment triggers" — specific conditions that say "if this happens, do that." End of quarter, the Strategist + EP read the triggers, decide, and the change goes into /02 — Strategy/ as the year-plan revision.
Send Jay (or post in #ai-workforce): the skill name, the input you gave, the wrong output, what you'd have written instead. Most "wrong" is missing context — usually a one-paragraph fix to the prompt.