05 — OUTPUT INTELLIGENCE: Advisory Onboarding Kit v2

Level 3 design. The kit produces a member kit with three operating modes. Each level builds on the previous one. Build artifacts that reflect this specific practice. Use the golden example as quality benchmark, not structural template. Fix what you can during production. Only escalate true judgment calls to Kathryn.

The Goal

Produce a complete member kit — a production system the practice owner uses to onboard clients. The kit has three operating modes. Levels are sequential: Level 1 is the foundation Level 2 and 3 need.

Level 1 Output — Structure (The Member Runs It)

Level 2 Output — Streamline (Member Manages, AI Produces)

Everything from Level 1, plus:

Level 3 Output — Delegate (AI Runs It, Member Oversees)

Everything from Levels 1 and 2, plus:

Required Inputs

1. Universal process (file 01) — the spine. Structure and design constraints.
2. AI extraction output (file 06) — the practice's specific data per offer.
3. Practice reference data — canonical names, tools, spellings.
4. Golden example (file 03) — Rob's roadmap. Quality benchmark.
5. Terminology lock (file 02) — words that mean specific things.

Production Order

Build sequentially. Each level requires the previous one.

Step 1 — Build the Onboarding SOP

Think of it as: The master document. If the member lost everything else, this alone would let them onboard a client.

What it must contain:

• Every step from trigger (sign + pay) through completion gate
• Who does each step (from the Q3 extraction data — advisor / mechanical / team member by name)
• When each step happens (timing relative to trigger)
• What each step produces (the deliverable or outcome)
• Decision points — where the member needs to make a judgment call vs. where the process is automatic
• Per-offer variations if the member has multiple offers

How to build it: Walk through the extraction's Q3 data (who touches what per step) and turn it into a narrative process document. Not a checklist — a readable SOP that someone could follow from start to finish.

Adapt to practice model:

• Lean/solo: Most steps are the advisor. The SOP emphasizes what to do and when, because there's no team to delegate to.
• Institutional: The SOP emphasizes handoffs — who passes what to whom, what the client sees vs. what happens internally.
• Coaching: The SOP follows activation cycles, not seasonal cadence.

Step 2 — Build the Client Deliverable Templates

Three documents, all populated with practice data.

The Roadmap (HTML)

Think of it as: The client opens this after signing and thinks "I know exactly what's happening and I feel good about this."

What it must do:

• Name the trigger (they're a client now — here's what that means)
• Show the first 7 days (welcome experience — concrete, not generic)
• Show the first 30 days (first win — dollar-measurable)
• Show the year (cadence — specific months and purposes)
• Show the boundary (what's included, what's separate)
• Name their people (team members and roles)
• Name the finish line (completion gate — criteria, not calendar)

What it must NOT do:

• Read like a contract (that's the mutual-responsibility doc)
• Include advisor-internal observations
• Use language the client wouldn't use
• Follow Rob's structure if this practice is fundamentally different

Adapt to the practice. A lean solo practice gets a simpler roadmap. An institutional practice gets more people and handoffs. A coaching practice gets activation cycles, not seasonal cadence. Don't force a practice into a structure that isn't theirs.

Brand: Dark theme (charcoal #1a1a1a, gold #b79d64, cream #f9f7f4), Cormorant Garamond + Inter. Hardcode hex — no var() CSS.

The Action Plan (Markdown)

Think of it as: The advisor's working doc per engagement. Pre-populated with practice data. Per-client specifics filled at kickoff.

What it must do:

• List what the firm is working on (with target dates)
• List what the client needs to do (with reasons — clients need to know WHY)
• Name the first win with specifics
• Name communication channels (who to contact for what)
• Preview the next meeting

Build from extraction data. If the first win is "identify $3K in overpaid estimated taxes," name that — not "identify savings opportunity."

The Mutual-Responsibility Doc (Markdown)

Think of it as: The handshake. What both sides commit to. Short, clear, signable.

What it must do:

• Name firm commitments (services, cadence, accessibility, scope)
• Name client commitments (attendance, documents, communication, tool usage)
• Name what's out of scope (from extraction's "what's separate")
• Name the completion gate
• Include signature lines

Step 3 — Build the Level 2 Kit Files

These turn the Level 1 deliverables into an AI-assisted production system.

Practice Context File

Think of it as: Everything Claude needs to know about this practice to produce onboarding deliverables. One file, readable, complete.

What it must contain:

• Practice name, advisor name, team roster with roles
• Each offer: name, what's included, what's separate, pricing structure (boundaries only)
• Per-offer: who touches what per onboarding step (the Q3 data)
• Tools the client interacts with
• Meeting cadence per offer
• First-win candidate per offer
• Welcome experience per offer
• Completion gate per offer
• Practice type

Build it from the extraction output. Reorganize for Claude's consumption — not a copy-paste of the extraction, but a clean reference document Claude can read and use.

AI Production Prompts

Think of it as: Step-by-step instructions the member copies to Claude to produce each deliverable.

What it must contain:

• A prompt for producing the roadmap for a new client
• A prompt for producing the action plan for a new client
• A prompt for producing the mutual-responsibility doc for a new client
• Each prompt references the practice context file and tells Claude what per-client inputs to use

Example shape:

## Produce a Roadmap

Load: [practice-context-file.md]
Reference: [onboarding-sop.md]

New client:
- Name: [Client Name]
- Package: [Offer Name]
- Start date: [Date]
- First meeting: [Date]

Produce the client roadmap. Use the practice context for all practice-specific details. Use the SOP for the step sequence. Fill in the client-specific fields above. Output as HTML matching the brand style.

Step 4 — Build the Level 3 Kit Instruction File

Think of it as: A CLAUDE.md for the member's practice. When their Claude instance loads this file, it knows how to onboard a client end-to-end.

What it must contain:

• Everything from the practice context file (embedded, not referenced — the file must be self-contained)
• The full SOP (embedded)
• Production instructions for all three client deliverables
• Quality checks (from file 04, adapted for this practice)
• The trigger: "When the member says 'onboard [name], [package], [date]' — produce the roadmap, action plan, and mutual-responsibility doc. Present for review."
• Escalation protocol: what to flag for the member's attention vs. what to resolve independently

This is the most important deliverable. It's what makes Level 3 work. A member with this file loaded in Claude can say "onboard Jane Smith, Full Advisory, starting July 1" and get all three documents back, populated and ready to review.

Build it as a real, working instruction file. Not a description of what it would do — an actual file the member can load.

Self-QC During Production

Check as you build. Don't produce everything and then run a quality pass.

Continuously verify:

• Every proper noun matches reference data
• Every practice-specific detail traces to extraction data
• Every client-facing section passes: "Would a client understand this cold?"
• The SOP and the client deliverables tell the same story (same steps, same people, same cadence)
• Level 2 prompts actually work — they reference real files and real data
• Level 3 instruction file is self-contained — no broken references

Fix issues immediately. Escalate only when you need Kathryn's judgment.

Cross-Member Intelligence

Use what worked for prior members. Don't copy proprietary content.

• If Rob's lean SOP structure worked, start there for similar practices
• If an institutional practice needed a different handoff section, build it from the start
• If a previous Level 3 instruction file had a gap, close it for this one

Delivery

Filename pattern per offer:

• [practice]-[offer]-onboarding-sop.md
• [practice]-[offer]-advisory-roadmap.html
• [practice]-[offer]-action-plan-template.md
• [practice]-[offer]-mutual-responsibility-template.md
• [practice]-[offer]-practice-context.md
• [practice]-[offer]-ai-prompts.md
• [practice]-[offer]-kit-instructions.md

Where it goes: The practice's per-member workspace — NOT the kit directory.

Production order per member: Extraction → SOP → client deliverables → practice context + prompts → kit instruction file. Sequential.

After Delivery — Mode 2 (Improve the Kit)

Every manual fix becomes a kit fix. If the member changes output by hand, if QC catches something, if the Level 3 instruction file needs adjustment — update the relevant kit file so the next run is better.

Cross-references

• File 01 — universal process (the spine)
• File 02 — locked vocabulary
• File 03 — golden example (quality benchmark)
• File 04 — quality framework
• File 06 — extraction conversation (source data)