Skip to main content

The Project Brief

In the "NanoClaw Hands-On Setup" lesson from "Meet Your First AI Employee - OpenClaw" chapter, you installed NanoClaw, connected it to WhatsApp, and designed a Layer 3 blueprint for your profession. You sketched out agent skills, MCP servers, and the domain expertise your AI employee would need. That blueprint has been sitting in your notes, waiting.

Now you build it. Over the next several lessons, you will turn that design into a working AI employee that responds to real messages, thinks like a domain expert, and handles professional tasks through WhatsApp. This is not a tutorial where you follow instructions step by step. Each lesson gives you a challenge, acceptance criteria, and hints if you get stuck. You bring your profession, your expertise, and your Layer 3 design.

The result is something you can actually use: an AI employee built for YOUR work, running on YOUR machine, accessible from YOUR phone.

What You Are Building​

Your AI employee is a NanoClaw instance configured for your specific profession. It receives messages via WhatsApp (and optionally other channels), processes them using Claude with your domain knowledge encoded as skills, and responds with profession-appropriate expertise.

By the end of the Bronze tier, your employee will have a professional identity, at least one domain skill, a working communication channel, and a logged conversation proving it works. Silver adds proactive behavior, trust boundaries, persistent memory, and a domain-specific report. Gold extends the architecture to multiple isolated groups.

Choose Your Tier​

TierPathTimeWhat You Build
BronzeThe Project Brief through Bronze Capstone: First Real Day~3 hoursIdentity + one skill + WhatsApp connection + proof of work
SilverThe Project Brief through Prove Professional Value~6 hoursBronze + scheduled tasks + permission boundaries + persistent memory + domain report
GoldThe Project Brief through Prove Professional Value (Gold track)~9 hoursSilver + multi-group architecture with per-group isolation

Pick the tier that matches your time and ambition. Every tier produces something functional. You can always return for the next tier later.

Your Blueprint​

Pull out your Layer 3 design from the "NanoClaw Hands-On Setup" lesson. This is your blueprint for everything that follows. It should contain:

  • 3+ Agent Skills you planned for your profession (the domain expertise to encode)
  • 3+ MCP Servers you identified (the external connections your employee needs)
  • Domain vocabulary and rules specific to your field

If your design is missing any of these, revisit it now and fill the gaps before moving forward.

Review your design and mark which skill you will implement first. Choose the one that would deliver the most value in a single conversation -- that is your Bronze target.

Acceptance Criteria​

  1. You have chosen a tier (Bronze, Silver, or Gold) and can explain why
  2. Your Layer 3 design from the NanoClaw Hands-On Setup lesson is reviewed, with your first skill identified
  3. Your GitHub repository exists with the structure below

Set Up Your Project Repository​

Create a repository called nanoclaw-employee on GitHub. Initialize it with this structure:

nanoclaw-employee/
  groups/
    global/
      CLAUDE.md           (shared rules across all groups)
    main/
      CLAUDE.md           (profession-specific identity)
  .claude/
    skills/
      (your skills will go here)
  README.md               (profession, tier choice, timeline)
  conversation-log.md     (you will fill this in Bronze Capstone: First Real Day)
  evaluation.md           (you will fill this in Bronze Capstone: First Real Day)

Your README should state your profession, your chosen tier, and when you plan to complete each lesson.

PRIMM-AI+ Practice​

This challenge follows the PRIMM-AI+ cycle. Before you build, predict. After you build, investigate.

Predict [AI-FREE]​

Before you start this challenge, write down:

  • What your completed repository structure will look like (list the folders and files you expect to create).
  • What your tier choice will be and the one sentence reason you will write in your README.
  • Whether your Layer 3 design will need any gaps filled before you can proceed.
  • Your confidence score from 1 to 5.

Do not ask the agent until those notes are written.

Run​

Use Claude Code ($ claude) to help you review your Layer 3 design, identify gaps, and scaffold the repository structure. Run git init and gh repo create from the terminal.

Complete the challenge as described above: choose your tier, review your Layer 3 design, and create the GitHub repository with the specified structure.

Investigate​

First, write your own explanation of whether the repository structure you created matches what you predicted. Then ask the agent: "Looking at my Layer 3 design, what gaps do you see between what I planned and what I would need to complete the Silver tier?"

Modify​

Change one requirement: swap your initial tier choice to the tier above or below it and rewrite the README rationale to justify the new choice. Predict how the project timeline section of your README will change, update it, and verify the README accurately reflects the new tier scope.

Make [Mastery Gate]​

Verify against the acceptance criteria above. Passing means: tier choice documented with rationale, Layer 3 design reviewed with first skill identified, and GitHub repository exists with the correct directory structure.

What does a completed AI employee look like across different professions? Here are four examples at each tier.

Accountant

  • Bronze: An employee that reviews invoice descriptions, flags missing fields, and checks tax line calculations via WhatsApp
  • Silver: Sends a daily summary of outstanding invoices every morning at 8am, refuses to file any document without CPA review
  • Gold: Separate groups for "Client A" and "Client B" with isolated financial data per client

Teacher

  • Bronze: An employee that drafts lesson activity ideas aligned to curriculum standards when you message it a topic
  • Silver: Sends weekly curriculum gap analysis every Sunday evening, flags any request to generate student grades
  • Gold: Separate groups for "Grade 5 Math" and "Grade 5 Science" with isolated planning contexts

Real Estate Agent

  • Bronze: An employee that researches comparable properties and summarizes listing details when you send an address
  • Silver: Monitors new listings daily and sends morning alerts for properties matching client criteria, refuses to state property values as facts
  • Gold: Separate groups for each client with isolated search histories and preferences

Freelance Designer

  • Bronze: An employee that tracks project briefs, extracts deliverable deadlines, and flags revision requests when you forward client messages
  • Silver: Sends a weekly capacity report every Monday, refuses to commit to timelines without checking current workload
  • Gold: Separate groups per client with isolated project contexts and communication logs