Home / Bizzby 101 / Module 3
Module 3 of 4

Module 3: How Your AI Team Works

Module 3: How Your AI Team Works

Course: Bizzby 101 — Working with Your AI Team Duration: ~35 minutes Lessons: 5 Outcome: You understand what AI agents can and can't do, how to give clear instructions, and how to review their work.

Lesson 3.1: What AI Agents Do (And Don't Do)

Format: 📖 Reading | Duration: 8 min

Hook

Let's get honest about what your AI team is great at, what it's okay at, and what it absolutely cannot do. Setting the right expectations now saves you frustration later.

Concept

What Your AI Agents Are Great At:

✓ Research. Competitor pricing, market trends, local demographics, best practices for your industry. Your agents can research in minutes what would take you hours or days.

✓ Writing. Marketing copy, social media posts, emails, website content, proposals, contracts, and scripts. Your agents write solid first drafts that you polish with your personal touch.

✓ Planning. Business plans, marketing calendars, launch timelines, campaign strategies, pricing models. Your agents think systematically and don't forget steps.

✓ Automation. Setting up booking confirmations, reminder sequences, follow-up emails, review requests. Once configured, these run on autopilot.

✓ Analysis. Reviewing your numbers, spotting trends, comparing your pricing to competitors, tracking your progress against goals.

✓ Organization. Keeping track of your tasks, deadlines, client information, and priorities. Your agents have perfect memory and never lose track of anything.

What Your AI Agents Are Okay At:

⚡ Creative work. Your agents can generate ideas and drafts, but breakthrough creative concepts usually come from you. Use your agents for execution, use your brain for the spark.

⚡ Advice. Your agents draw on broad business knowledge, but they don't know your specific neighborhood, your personality, or your gut instinct. Treat agent advice as informed input, not gospel.

⚡ Nuanced communication. Your agents write well, but your personal voice matters for client relationships. Use agent drafts as starting points and add your personality.

What Your AI Agents Cannot Do:

❌ Physical work. They can't clean a house, fix a sink, or deliver a package. You (or someone you hire) handle service delivery.

❌ Make phone calls. Your agents can draft what to say, but you make the calls. (Voice AI integration is on our roadmap.)

❌ Access your bank accounts. Financial tracking is based on numbers you share. Your agents don't have access to your bank, credit card, or payment processor accounts.

❌ Make final decisions. Your agents recommend. You decide. Nothing goes live without your approval.

❌ Replace human relationships. Client relationships, networking, partnerships, and community involvement require you. Your agents support these activities but can't replace your presence.

The Human-in-the-Loop Model

Bizzby uses a "human-in-the-loop" approach. This means:
  1. You set direction. Tell your team what you want to accomplish.
  2. Agents do the work. Research, write, plan, build, analyze.
  3. You review and approve. Nothing goes public without your sign-off.
  4. Agents adjust. Based on your feedback, they refine and improve.
  5. You execute. Show up for clients, deliver your service, build relationships.
This model works because it combines what AI is best at (speed, scale, consistency, memory) with what humans are best at (judgment, creativity, relationships, physical service delivery).

Example

Rachel started a pet sitting business. In her first week, here's how the human-in-the-loop model worked:

Quiz

  1. Which of these can your AI agents do?
- a) Walk dogs for clients - b) Research competitor pricing in your area ✓ - c) Access your bank account - d) Make phone calls to leads
  1. What does "human-in-the-loop" mean?
- a) A human builds the AI system - b) You review and approve all agent work before it goes live ✓ - c) You do all the work and agents watch - d) AI makes all decisions automatically

Action Item

Think about your business and make two lists:
  1. Things I want my agents to handle: (marketing, scheduling, writing, research, etc.)
  2. Things only I can do: (client delivery, phone calls, networking, etc.)
This mental exercise helps you delegate effectively from Day 1.

Lesson 3.2: Watching Alex Work: A Day in the Life

Format: 🎬 Video | Duration: 12 min

Video Description

Screen recording showing a full day of agent activity for a real business. Timestamps show morning briefing, mid-day task completion, approval workflow, and end-of-day summary. Demonstrates the behind-the-scenes coordination between agents.

Script Outline

  1. 7 AM - Morning Briefing (2 min): Alex sends the daily summary. Show what it looks like. Three items need approval, two tasks completed overnight.
  2. 8 AM - Founder Reviews (2 min): Founder opens dashboard. Approves a social post, gives feedback on an email draft ("make it shorter and more casual"), declines a pricing suggestion ("too aggressive for my market").
  3. 10 AM - Agents Execute (3 min): Show Maya revising the email based on feedback. Sam updating the website with a new testimonial. Kai running the automated follow-up sequence for yesterday's new inquiry.
  4. 2 PM - New Task (2 min): Founder messages Alex: "A competitor just dropped their prices by 20%. What should we do?" Show Alex coordinating Maya (competitive analysis) and Riley (margin impact) to deliver a recommendation within an hour.
  5. 5 PM - End of Day (3 min): Activity feed summary. Founder sees everything that happened. Sends Alex a quick "good day, thanks" message.

Supporting Text

A typical day looks like this:

Your agents don't sit idle waiting for instructions. They have a running task list based on your goals and priorities. Here's what a typical day looks like behind the scenes:

Morning (before you wake up): Midday: Evening: You only see what matters to you. The behind-the-scenes coordination happens automatically. You focus on approvals, feedback, and client work.

Action Item

No action required. Just understand the flow. Your agents are already working based on the information you shared in Module 2.

Lesson 3.3: How to Give Great Instructions

Format: 📖 Reading | Duration: 5 min

Hook

The quality of your team's work is directly tied to the quality of your instructions. Good input, great output. Vague input, mediocre output. Here's how to be a great boss to your AI team.

Concept

The CCC Framework: Clarity, Context, Constraints

Every time you ask your team to do something, include three things:

1. Clarity — What do you want? Be specific about the outcome you're looking for.
Vague ❌Clear ✓
"Write me some social media posts""Write 5 Instagram posts promoting my lawn care spring cleanup special ($99 for first visit)"
"Help me with pricing""Research what residential cleaning businesses in Portland charge for a 3-bedroom deep clean"
"Make me a website""Create a simple 3-page website: home page with my services, about page, and a booking page"
2. Context — What should they know? Share relevant background that shapes the work.
No Context ❌With Context ✓
"Write an email to my clients""Write an email to my 12 recurring cleaning clients letting them know I'm raising prices by $10/visit starting March 1. Keep it warm and grateful."
"Create a marketing plan""Create a marketing plan for my pressure washing business. I'm in suburban Dallas, targeting homeowners with houses 10+ years old. My budget is $200/month."
3. Constraints — What are the limits? Tell your team about any boundaries.
No Constraints ❌With Constraints ✓
"Write some ad copy""Write Google ad copy for my tutoring service. Max 30 characters for headline, 90 for description. Target parents of high school students."
"Build me a social media presence""Focus on Instagram and Nextdoor only. I don't want to use TikTok or Twitter. Post 3 times per week max."

Quick Reference

Before sending a task to Alex, check:

Example

Before CCC:
"Alex, I need help with marketing."
After CCC:
"Alex, I need 5 Nextdoor posts for my dog walking business in the Buckhead neighborhood of Atlanta. I charge $25 per 30-minute walk. Tone should be friendly and local. I don't want any posts that mention competitor apps like Rover or Wag. Schedule them for Monday through Friday this week."

The second message gives your team everything they need to deliver great work on the first try.

Action Item

Take one task you've been thinking about and write it using the CCC framework. Send it to Alex. Notice the difference in the quality of the response compared to a vague request.

Lesson 3.4: Reviewing Agent Output

Format: 📖 Reading | Duration: 5 min

Hook

Your agents are fast and thorough, but they're not mind readers. Giving good feedback makes them better every single time.

Concept

When your agent delivers work for review, you have four options:

1. Approve ✓ "This is good. Go ahead." Use this when the work meets your standards and is ready to go. 2. Approve with Minor Edits ✏️ "This is 90% there. Change X and then publish." Use this when the work is close but needs a small tweak. Example: "Love the post, but change 'premier' to 'friendly neighborhood.' Less corporate, more personal." 3. Send Back for Revision 🔄 "This misses the mark. Here's what I want instead." Use this when the direction is wrong. Be specific about what to change and why. Example: "This email is too salesy. I want it to feel like a personal note from me, not a marketing blast. Rewrite it in a conversational tone, like I'm texting a neighbor." 4. Decline ❌ "I don't want to do this." Use this when you disagree with the approach entirely. Example: "I don't want to run Facebook ads right now. Let's focus on organic Nextdoor posts instead." Feedback That Helps:
Unhelpful ❌Helpful ✓
"I don't like this""The tone is too formal. Make it casual and friendly."
"This isn't right""The pricing is too low. Competitors charge $150-$200. Adjust to $175."
"Try again""Good structure, but I want more emphasis on our eco-friendly products. That's our main selling point."
The Feedback Loop: The more feedback you give, the better your agents get at matching your preferences. After a few weeks, you'll notice that first drafts start hitting closer to what you want. Your agents learn your style, your voice, and your standards.

Example

Tom runs a junk removal business. Maya drafted a Facebook ad: Draft 1: "Professional junk removal services. Licensed and insured. Call today for a free estimate!" Tom's feedback: "Too generic. Everyone says this. I want to lead with the fact that we recycle and donate 70% of what we haul. That's what makes us different. Make it feel like we care about the environment, not just hauling trash." Draft 2: "We don't just haul your junk. We find it a new home. 70% of what we remove gets recycled or donated. Your old stuff deserves better than a landfill. Book a pickup today." Tom: "That's us. Approved."

Action Item

The next time an agent sends you work to review, practice using specific feedback. Instead of "looks good" or "not quite," explain what you'd change and why. Your agents get better fast.

Lesson 3.5: Your First Agent Task

Format: 🎯 Action | Duration: 10 min

Hook

You've learned how your team works. You know how to give instructions and review output. Time to put it into practice with a real task.

What to Do

Pick ONE task from the list below (or create your own) and send it to Alex using the CCC framework. Task Options: If you're just starting out: If you already have some clients: If you're growing:

How to Send It

Use the CCC framework:
"Alex, here's my first task: What I want: [specific outcome] Context: [relevant background] Constraints: [any limits or preferences] Please have this ready for my review by [timeframe]."

What to Expect

Alex will acknowledge your task, ask any clarifying questions, and delegate to the right agent. Expect a first draft within a few hours (often much faster). Review it using the four-option framework from Lesson 3.4.

Action Item

Send the task now. Don't overthink it. The goal is to experience the workflow, not to pick the perfect task.

Module 3 Quiz

4 questions — instruction quality, output review, agent capabilities
  1. What does the CCC framework stand for?
- a) Copy, Create, Complete - b) Clarity, Context, Constraints ✓ - c) Client, Customer, Conversion - d) Communication, Coordination, Control
  1. Your agent sends you a social media post. The tone is too formal. What do you do?
- a) Approve it anyway - b) Delete it and write your own - c) Send it back with specific feedback about the tone you want ✓ - d) Ignore it
  1. Which of these is something your AI agents CANNOT do?
- a) Write marketing emails - b) Research competitor pricing - c) Make phone calls to potential clients ✓ - d) Create a booking automation
  1. Why is giving feedback important?
- a) It's required by Bizzby - b) It helps your agents learn your preferences and improve over time ✓ - c) It generates more charges - d) It's optional and doesn't matter

Module 3 Agent Prompt

After completing all lessons, send this to Alex:

"Alex, please create a brief summary of what you know about my business so far and list the top 3 things you plan to work on this week."