bni-chapter-be-ai-first-implementation-protocol

name: BNI Chapter — Be AI-First Implementation Protocol description: BNI-domain specialisation of the Be AI-First framework. 4-phase sequence (Learn → Wire → Automate → Scale + Step 0 Discovery + Governance) for transforming a BNI chapter's weekly operations using the bni121 platform as the reference implementation. Each phase has BNI-specific deliverables, gates, and agent menus. Invoke when onboarding a new chapter or scoping an automation engagement for an existing chapter. type: pattern tags: [bni, bni121, chapter-ops, ai-first, supabase, drip-worker, gcal, zoom, referrals, palms, hopetech]

BNI Chapter — Be AI-First Implementation Protocol

Sister skill to be-ai-first-framework. That one is generic across sectors; this one is BNI-specific — every example, agent, and gate is anchored to chapter weekly ops and the bni121 platform.

Pre-existing harvested skill bni-crm-build-pitfalls covers technical gotchas while building the platform — read both together when onboarding a new chapter.

Problem

BNI chapter Leadership Teams (LT) rotate annually and inherit a tangle of WhatsApp groups, paper visitor slips, manual member-trackers, and screenshots of attendance. Symptoms we see in every chapter:

• Visitors never get a follow-up — chapter loses 60–80% of conversions
• 1-2-1s happen but get logged days later, if at all — PALMS scores rot
• Referrals get spoken in the meeting but never written down — credit lost
• The new LT spends Q1 rebuilding spreadsheets the previous LT also rebuilt
• "AI" in the chapter = one member using ChatGPT for marketing copy, nobody else

The Be AI-First framework, applied with bni121 as the platform, fixes this in a 10-week sprint and leaves the chapter with a reusable Business Brain that survives LT rotation.

BNI-specific phase map

Step 0 — Pre-Engagement Discovery (BNI version)

Goal: Score the chapter's current state and pick a Champion before Day 1.

BNI tool & workflow audit

Tasks

• Pull member roster from BNI Connect → CSV
• List every tool the LT uses (BNI Connect, WhatsApp groups, Zoom, Google Calendar, Excel, Drive)
• Score AI literacy of LT (President / VP / Secretary-Treasurer / Membership / Visitor Host / Education) on 0–5
• Identify the AI Champion — usually Education Coordinator or Secretary-Treasurer (curious, weekly-cadence role)
• Note cultural resistance: any older members who'll push back on automation? Plan for them.

Gate (must pass before Phase 1)

• Member roster exported as CSV
• Tool inventory complete (≥6 tools listed with active-user counts)
• AI Champion named, agreed by President + VP
• Top 5 chapter pain-points documented (visitor follow-up, 1-2-1 logging, etc.)

Deliverable: Chapter Readiness Scorecard — one page, signed by President.

Phase 1 — LEARN (Weeks 1–2)

Goal: LT Champion + 1–2 secondary Champions hit personal AI fluency on real chapter work.

Daily practice (Champions)

• 1 hour/day on Claude.ai or Claude Code, doing actual chapter tasks:
  • Draft this week's visitor follow-up emails (real names from last meeting)
  • Summarise last 4 weeks of meeting minutes into a member newsletter
  • Generate 1-2-1 questions tailored to a member's profession
  • Draft a substitute-finder WhatsApp message
  • Convert PALMS spreadsheet into a chapter health summary
• Log 3 "aha moments" per week — tasks where AI saved 30+ minutes of LT time

Weekly share-back

• 30-min slot at the LT meeting (not the chapter meeting): each Champion demos one win
• Consolidate into a chapter What AI Can Do for BNI doc

Deploy ironbark early

• Every Champion session auto-harvests reusable prompts
• By end of Phase 1, the chapter has a starter prompt library — these become Phase 3 agents

Gate

• Each Champion has 10+ hours hands-on
• Capability Map written: "AI handles this well" (drafts, summaries, follow-ups) vs. "AI needs a human" (induction speeches, member resignations, conflict)
• President signs off on Capability Map

Deliverable: BNI Capability Map — 2-column doc with realistic per-task time savings (e.g., "visitor follow-up: 15 min → 2 min").

Phase 2 — WIRE (Weeks 3–5)

Goal: Stand up the chapter's Business Brain. Reference implementation = bni121.

Knowledge structuring (markdown-first)

Convert these to markdown / Notion. No Word/PDF as primary source.

• BNI core values & code of ethics (chapter-adapted)
• Weekly meeting agenda template
• Visitor-host script
• Member induction SOP
• Substitute-finder SOP
• Referral handoff SOP
• 1-2-1 question bank (by profession category)
• Education-slot library
• Chapter org chart with LT roles, decision authority, term dates

Live data connections (the bni121 wiring pattern)

Wire all of the above into one Supabase project with multi-tenant scoping (one tenant per chapter). Reference the harvested Auto-Tenant Fetch Wrapper pattern so every API call carries tenant_id automatically — see bni-crm-build-pitfalls for the gotchas.

Single source of truth (per data type)

• Members → Supabase members (NOT BNI Connect, NOT WhatsApp, NOT Excel)
• Schedule → Google Calendar (mirrored to Supabase by gcal_worker.py)
• Templates → Supabase templates (NOT individual member's drafts folder)

Gate

• Chapter Context File written (1–2 pages) covering chapter mission, LT roles, weekly cadence
• At least 3 live data sources connected (Supabase + Google Calendar + Zoom is the minimum)
• All SOPs in markdown
• LT can log into the bni121 tenant and see their real members

Deliverable: A working Chapter Business Brain — bni121 deployment with real data flowing.

Phase 3 — AUTOMATE (Weeks 6–10)

Goal: Ship 6+ BNI agents to production, each with a scoring checklist and human-in-loop gate.

BNI agent menu (priority order)

Closed-loop rules (apply to ALL agents)

• Output passes a scoring checklist before send / action — see harvested Zero-Hardcode Bottom-Up Integration Protocol
• Agent logs each iteration into Supabase: input, output, who-approved, what-changed
• Quality threshold per agent (e.g., visitor email must include: visitor's first name, member who invited, specific business benefit, soft CTA — score < 4/4 → loop back)
• LT weekly 15-min retro: which agents earned trust, which need prompt refinement
• All revised prompts stored in Supabase prompt_library (also harvested by ironbark to ~/.claude/skills/)

Gate

• ≥1 live agent in each LT role's domain (Visitor Host, Membership, Secretary-Treasurer have agents)
• Every agent has a documented scoring checklist in markdown
• Closed-loop review = standing 15-min on weekly LT meeting
• At least 4 weeks of agent output history in Supabase for audit

Deliverable: 6+ live agents + chapter Skills Library v1.

Phase 4 — SCALE (Week 11+)

Goal: Replace LT manual reporting with an Intelligence Layer. Track chapter performance by leverage, not effort.

Intelligence layer (replaces LT busywork)

• Daily 8am brief to LT WhatsApp/email: yesterday's referrals, today's 1-2-1s, this week's visitors, members at risk
• Weekly chapter health dashboard — President sees exceptions only; auto-summary of attendance %, PALMS, traffic lights
• Meeting minutes → tasks pipeline: Zoom recording → transcript → action items → assigned in tracker.html
• Recurring member questions ("when's the next visitor day?", "what's the dress code?") → routed to Business Brain → LT only handles novel ones

Chapter-specific token-max metrics

• Chapter AI Leverage Ratio = tasks completed per week / hours of LT input
  • Quarterly target: 2× → 5× → 10×
• LT Time Saved per Week = before-vs-after stopwatch on the 10 audited workflows
• Conversion ratio improvement = visitors → returning visitors → applications → inducted members (track each gate, attribute lift to specific agents)
• PALMS lift per member = average score improvement since AI agents went live

Continuous expansion

• Re-run Learn → Wire → Automate on the next 10 BNI workflows: testimonials capture, Power-Team meetings, mentor matching, induction-speech drafting, anniversary tracking, etc.
• Build out the prompt_library — every refined prompt is reusable across BNI chapters using bni121
• Annual LT rotation handover = export Skills Library + Business Brain → next LT inherits everything

Gate

• Chapter AI Leverage Ratio tracked and improving QoQ
• Daily intelligence brief auto-sent and reviewed by LT
• Skills Library has ≥20 reusable prompts across all 6+ agents

Governance & Change Management (BNI specifics)

Roles

• AI Ops Lead = the Champion identified in Step 0 (typically Education Coordinator). Owns the Business Brain, Skills Library, and quality scores.
• Human-in-loop reviewer per agent type:
  • Visitor outputs → Visitor Host
  • Referral nudges → giver and receiver members
  • PALMS / health flags → Membership Coordinator
  • Public/external messages → President signs off until trust ≥ 9/10
• LT rotation handoff: outgoing AI Ops Lead does a 1-hour transfer with incoming, walks through Skills Library, agent log history, and prompt-library access.

Recurring rituals

• Weekly 15-min AI Retro — embed inside the existing LT meeting, not a new meeting
• Monthly prompt audit — any prompt with < 80% approval rate gets refined or retired
• Quarterly 90-day review with the chapter at large — share Leverage Ratio + time saved (members will renew faster when they see numbers)

Data privacy & safety (HARD rules — BNI is high-trust)

• Member contact data classification: internal-only by default; never paste into external AI APIs without an approved tool list
• Visitor data: treat as "lead data under DPA" — never share across chapters without consent
• Referral content: reviewed by giver before any agent-drafted outreach is sent on their behalf
• Recordings (Zoom): transcripts stay in Supabase; no third-party speech APIs without LT approval
• Approved AI tool list maintained in Supabase approved_tools — adding a tool requires LT vote

Slash command form

Drop these into ~/.claude/commands/ for one-line invocation. Pattern stolen from harvested Custom Slash Command — Encode Multi-Step Protocols.

# ~/.claude/commands/bni-discover.md

We are kicking off a Be AI-First engagement for chapter: **$ARGUMENTS**

Run Step 0 (BNI version). Output ONLY:

1. Chapter Readiness Scorecard template (markdown)
2. Tool inventory CSV (columns: tool, monthly_cost, active_users, ai_ready, status)
3. LT AI literacy scoring sheet (rows: President, VP, Secretary-Treasurer, Membership, Visitor Host, Education, Mentor)
4. AI Champion shortlist (criteria: weekly cadence, curious, owns data already)
5. Top-5 chapter pain-points capture form

Do NOT propose agents yet. Do NOT design Supabase schema yet. Wait for the LT to fill the templates.

Start now.

# ~/.claude/commands/bni-wire.md

We are starting Phase 2 (WIRE) for BNI chapter: **$ARGUMENTS**

Pre-condition: confirm Phase 1 cleared (Capability Map signed by President). If not, stop and tell user to run /bni-learn first.

Then propose, in this order:
1. SOP markdown migration list (which BNI SOPs, which order)
2. Chapter Context File draft (1–2 pages)
3. bni121 tenant deployment plan (Supabase project, RLS policies, seed data)
4. The 3 live data sources to connect first (default: Supabase members + Google Calendar + Zoom)

When done, say: "Phase 2 plan complete — awaiting LT sign-off."

# ~/.claude/commands/bni-agent.md

Build a new BNI agent for chapter: **$ARGUMENTS**

Follow the closed-loop rules:
1. State the agent's job in one sentence.
2. List its data inputs (Supabase tables / Google Calendar / Zoom transcripts).
3. Define the scoring checklist (4–6 binary criteria).
4. Define the human-in-loop reviewer (which LT role).
5. Define the failure mode (what counts as score < threshold → loop back).
6. Implement the agent (Python worker pattern: see drip_worker.py / gcal_worker.py).
7. Add a DATA SOURCE traceability comment.

Do NOT ship without the scoring checklist.

Add similar /bni-learn, /bni-automate, /bni-scale, /bni-review commands as the chapter scales.

Reference implementation — the bni121 codebase

PWA install — manifest.json + sw.js already wired; chapter members can add to home screen.

Engagement packages (chapter pricing)

DrmHope's "Enterprise software. Shipped in 6 weeks." promise applies — fixed scope, fixed timeline.

When to invoke this skill

• Onboarding a new BNI chapter to bni121
• Starting an AI-First engagement with an existing chapter
• LT rotation: incoming team needs to inherit prompt library + agent ownership
• Quoting a BNI region (multiple chapters at once) — same skill, applied per chapter
• Auditing an in-flight chapter rollout — walk the test harness for the current phase

When NOT to invoke

• Generic non-BNI client engagement → use be-ai-first-framework
• One-off bni121 platform bug → use the harvested bni-crm-build-pitfalls instead
• LT just wants ChatGPT licences without the framework → say no; framework is what makes it stick

Why it works for BNI specifically

Source documents

• Client-facing PDF (12 pages): ~/Desktop/HopeTech_AI_First_Framework.pdf
• PDF generator (re-runnable): 1backup/BNI 121/build_aifirst_guide.py
• Generic framework skill: .claude/skills/be-ai-first-framework/SKILL.md
• Build-time gotchas: .claude/skills/harvested/bni-crm-build-pitfalls/SKILL.md
• Original framework reference: ~/Downloads/be_ai_first_guide.pdf
• Live deployment: https://bni121.vercel.app